![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F443651%2F67ca203500555bdbcb5434995031ea88%2FScreenshot%20from%202020-08-14%2012-59-35.png?generation=1597402853523722&alt=media\")
Am I being overly skeptical or is just plain weird? Notice the very low submission counts, the fact that they are all still novice, similar latest submission dates and especially almost the same scores.
\n
EDIT: There are many more than these 7 accounts involved. I find at least 10 to be very suspicious. Some other clear indicators of private sharing can also be found for other \"clusters\" of teams.
\nEDIT2: I got contacted by an anonymous source who shared very valuable information regarding this situation. Please see my topic here.
", "messages": [ { "id": 970315, "postDate": "2020-08-14T11:00:55.377Z", "content": "Am I being overly skeptical or is just plain weird? Notice the very low submission counts, the fact that they are all still novice, similar latest submission dates and especially almost the same scores.
\n
EDIT: There are many more than these 7 accounts involved. I find at least 10 to be very suspicious. Some other clear indicators of private sharing can also be found for other \"clusters\" of teams.
\nEDIT2: I got contacted by an anonymous source who shared very valuable information regarding this situation. Please see my topic here.
", "rawMarkdown": "Am I being overly skeptical or is just plain weird? Notice the very low submission counts, the fact that they are all still novice, similar latest submission dates and especially almost the same scores.\n\n\n\n**EDIT**: There are many more than these 7 accounts involved. I find at least 10 to be very suspicious. Some other clear indicators of private sharing can also be found for other \"clusters\" of teams. \n\n**EDIT2**: I got contacted by an anonymous source who shared very valuable information regarding this situation. [Please see my topic here.](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175156)\n\n", "votes": 122 }, { "id": 971492, "postDate": "2020-08-15T15:21:39.683Z", "content": "What a strange coincidence. All 7 of these teams jumped over me today. How can 7 teams working independently all make the same discovery (to increase their score) at the same time?
\n(Note that this happened in a span of a few hours and no high scoring public notebooks were published during this time).
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1723677%2F87fc24b01cc9f9d08f145ac87c74c4d8%2Fjump.png?generation=1597504831812559&alt=media\")
In the span of (maximum) 2 hours even :). What a coincidence indeed!
", "rawMarkdown": "In the span of (maximum) 2 hours even :). What a coincidence indeed!", "votes": 3 }, { "id": 971496, "postDate": "2020-08-15T15:27:36.737Z", "content": "@cdeotte I'm pretty new to these type of antics, what are they trying to achieve?
", "rawMarkdown": "@cdeotte I'm pretty new to these type of antics, what are they trying to achieve?", "votes": 5 }, { "id": 971501, "postDate": "2020-08-15T15:37:08.360Z", "content": "Obviously group behavior of the same organization
", "rawMarkdown": "Obviously group behavior of the same organization ", "votes": 1 }, { "id": 971512, "postDate": "2020-08-15T15:59:37.670Z", "content": "Coïncidence ? Je ne crois pas… ^^
", "rawMarkdown": "Coïncidence ? Je ne crois pas... ^^ ", "votes": 8 }, { "id": 971523, "postDate": "2020-08-15T16:17:35.123Z", "content": "Expecting this is not gonna get noticed in the heart of the data science community, oh the irony…
", "rawMarkdown": "Expecting this is not gonna get noticed in the heart of the data science community, oh the irony...", "votes": 14 }, { "id": 971661, "postDate": "2020-08-15T19:28:16.657Z", "content": "@reighns They could be trying to sell accounts? I don't see any other way to benefit off of having multiple accounts with nearly identical placements…
", "rawMarkdown": "@reighns They could be trying to sell accounts? I don't see any other way to benefit off of having multiple accounts with nearly identical placements... ", "votes": 1 }, { "id": 971711, "postDate": "2020-08-15T20:23:14.233Z", "content": "the Wang Cheng Yang cabal strikes again! People have put their soul into this comp, so unfair if there is any cheating or collusion going on. Private sharing is unfair compared to a legitimate team. Discouraging. Imagine being that person that just misses a gold, silver or bronze by one position.
", "rawMarkdown": "the Wang Cheng Yang cabal strikes again! People have put their soul into this comp, so unfair if there is any cheating or collusion going on. Private sharing is unfair compared to a legitimate team. Discouraging. Imagine being that person that just misses a gold, silver or bronze by one position.", "votes": 8 }, { "id": 971804, "postDate": "2020-08-15T22:59:56.560Z", "content": "@cdeotte, since I discovered something called \"click fraud \" when I participated in the TalkingData AdTracking Fraud Detection Challenge, my distrust level for this kind of thing went up exponentially.
\nI agree with @serigne. Grosse coïncidence en effet. J'en doute.
", "rawMarkdown": "@cdeotte, since I discovered something called \"click fraud \" when I participated in the [TalkingData AdTracking Fraud Detection Challenge](https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection), my distrust level for this kind of thing went up exponentially.\n\nI agree with @serigne. Grosse coïncidence en effet. J'en doute.", "votes": 3 }, { "id": 972089, "postDate": "2020-08-16T08:21:10.730Z", "content": "@maggiemd, @philculliton, wouldn't this be worth a look at? There are several cases of Novice and not-Novice jumps probably using the same non-public submissions. But this specific team really is … an outlier considering the attribute \"audacity\"
\nThe irony of fraud detection not working in the world's best Data-Science community is almost worth an own notebook 😃
I am so glad you posted this Chris… I mean some people, including myself, work really hard alone (or in a team of 2) to get a medal and here comes this mafia stealing from us… I'm always 10 spots behind a medal, I really wonder why….. Thank you for raising this questions, maybe Kaggle staff wakes up….
", "rawMarkdown": "I am so glad you posted this Chris... I mean some people, including myself, work really hard alone (or in a team of 2) to get a medal and here comes this mafia stealing from us... I'm always 10 spots behind a medal, I really wonder why..... Thank you for raising this questions, maybe Kaggle staff wakes up....", "votes": 1 } ] }, { "id": 970515, "postDate": "2020-08-14T14:21:47.183Z", "content": "Some similarities in when subs were made too.
\nThis shows subs from just the last 5 days. Only shows subs that increased LB score. Makes sense that a group would sub at the same time of day if they are coordinating… or could just be a coincidence.
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F644036%2F1951f6a1088157c5563f28918b309c3e%2Fsubs.png?generation=1597414585360009&alt=media\")
Thanks @robikscube, that's indeed a great indicator. It's insane how obvious they have made it…
\nThree of them are also collaborators on the same dataset: https://www.kaggle.com/wangxueqing618/tfrecord384x384
", "rawMarkdown": "Thanks @robikscube, that's indeed a great indicator. It's insane how obvious they have made it...\n\nThree of them are also collaborators on the same dataset: https://www.kaggle.com/wangxueqing618/tfrecord384x384", "votes": 11 }, { "id": 970775, "postDate": "2020-08-14T19:21:04.980Z", "content": "Wow.It explains everything.
", "rawMarkdown": "Wow.It explains everything.", "votes": 2 }, { "id": 970925, "postDate": "2020-08-15T02:27:08.827Z", "content": "@robikscube wow, I'm convinced. I am sure Kaggle does basic things like check for accesses from the same IP address. Most people are smart enough to use VPN though. Kaggle should have its own Kaggle competition to detect fraud, and then use what it learns to improve.
\nOne basic way would be to charge………..I realize that has all kinds of problems, but it does stop people from creating basic farm accounts en masse. Another way would be to do basic things like cell # verification. I realize people can have multiple #'s or google voice, etc. But I mean, if you put enough controls in place it does help things.
", "rawMarkdown": "@robikscube wow, I'm convinced. I am sure Kaggle does basic things like check for accesses from the same IP address. Most people are smart enough to use VPN though. Kaggle should have its own Kaggle competition to detect fraud, and then use what it learns to improve.\n\nOne basic way would be to charge...........I realize that has all kinds of problems, but it does stop people from creating basic farm accounts en masse. Another way would be to do basic things like cell # verification. I realize people can have multiple #'s or google voice, etc. But I mean, if you put enough controls in place it does help things. ", "votes": 5 }, { "id": 977003, "postDate": "2020-08-19T08:20:02.537Z", "content": "Out of curiosity, how do you get the data to create this plot?
", "rawMarkdown": "Out of curiosity, how do you get the data to create this plot? \n\n" }, { "id": 977382, "postDate": "2020-08-19T12:53:51.257Z", "content": "@janidziak, you look at the public leaderboard tab, there is a \"download raw data\" button you can click to get a CSV file of the submissions. It shows all submissions that caused teams to increase their public leaderboard score.
", "rawMarkdown": "@janidziak, you look at the public leaderboard tab, there is a \"download raw data\" button you can click to get a CSV file of the submissions. It shows all submissions that caused teams to increase their public leaderboard score.", "votes": 1 }, { "id": 979115, "postDate": "2020-08-20T16:16:38.357Z", "content": "Thank you @robikscube
", "rawMarkdown": "Thank you @robikscube " } ] }, { "id": 972076, "postDate": "2020-08-16T07:59:08.897Z", "content": "I wish those are just more accounts for GPU/TPU time. But kaggle's reputation has been downfalling fast in my country because there are so called medal factories.
\nA medal factory sold medals by sharing private notebooks. Thousands of dollars for each one getting a medal with their private notebook in the cost of 20yuan/3dollars for VPN each month. Gold medals worth more. And people buying those medals will use kaggle's reputation and medal quality in trade for jobs/applying graduates in school etc. They damage kaggle's reputation, got their own benefit and then leave, possibly never come back.
\nMy teammate and I have devoted hours and days in this competition to improve our cv and climp the LB little by little. All of a sudden those novices just come up and surpass you. When the legend of medal factories happened, things can be very FRUSTRATED and MIND-FFFFING. SAD indeed. Hope kaggle will not be another \"game destroyed by cheaters\". I love this community.
\nI wish legend is just legend but not truth.
\n@philculliton , @juliaelliott, @Phil Culliton
", "rawMarkdown": "I wish those are just more accounts for GPU/TPU time. But kaggle's reputation has been downfalling fast in my country because there are so called medal factories.\n\nA medal factory sold medals by sharing private notebooks. Thousands of dollars for each one getting a medal with their private notebook in the cost of 20yuan/3dollars for VPN each month. Gold medals worth more. And people buying those medals will use kaggle's reputation and medal quality in trade for jobs/applying graduates in school etc. They damage kaggle's reputation, got their own benefit and then leave, possibly never come back. \n\nMy teammate and I have devoted hours and days in this competition to improve our cv and climp the LB little by little. All of a sudden those novices just come up and surpass you. When the legend of medal factories happened, things can be very FRUSTRATED and MIND-FFFFING. SAD indeed. Hope kaggle will not be another \"game destroyed by cheaters\". I love this community.\n\nI wish legend is just legend but not truth.\n\n@philculliton , @juliaelliott, @Phil Culliton\n", "votes": 28 }, { "id": 972601, "postDate": "2020-08-16T17:12:00.580Z", "content": "And this team is just LEGENDARY. (all screenshots are from public LB)
\nLiverpool:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2F20f0f6640088f38d6edefbe4f6f9db0e%2Fliverpool.png?generation=1597597796074934&alt=media\")
Alaska2:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2Fc7d82776d48187056c13df3680f25cea%2Falaska.png?generation=1597597815980223&alt=media\")
Panda:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2F73878e8b785a15ce582752c7be22037c%2Fpanda.png?generation=1597597833335107&alt=media\")
Wheat:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2Fcefcfc211bcc5ace3e34a191f90c8557%2Fwheat.png?generation=1597597850061151&alt=media\")
Melanoma:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2F3d96933063d421d50317d7c7f540fe5c%2Fmelanoma.png?generation=1597597864429626&alt=media\")
Landmark2020:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2F8d3203bdb886ebc5b67081bb58b83fa8%2Flandmark2020.png?generation=1597597879503901&alt=media\")
Yet Kaggle has done nothing to them. Is there any explaination?
", "rawMarkdown": "And this team is just **LEGENDARY**. (all screenshots are from public LB)\nLiverpool:\n\n\nAlaska2:\n\n\nPanda:\n\n\nWheat:\n\n\nMelanoma:\n\n\nLandmark2020:\n\n\nYet Kaggle has done nothing to them. Is there any explaination?", "votes": 19, "replies": [ { "id": 972609, "postDate": "2020-08-16T17:18:16.753Z", "content": "That makes some team above them suspicious 🙂
", "rawMarkdown": "That makes some team above them suspicious 🙂 ", "replies": [ { "id": 973230, "postDate": "2020-08-17T07:26:18.983Z", "content": "@jielu0728 same profile as all the accounts I found and others posted.
\nNo activity, no posts, no comments, no bio, no followers, not following anyone. They just joined Kaggle and started climbing competitions. So blatant, in your face……..I mean no shame at all.
\nTop 1% in Wheat and Melanoma…………..about to take down medals if Kaggle does not do something about it.
", "rawMarkdown": "@jielu0728 same profile as all the accounts I found and others posted.\n\nNo activity, no posts, no comments, no bio, no followers, not following anyone. They just joined Kaggle and started climbing competitions. So blatant, in your face........I mean no shame at all. \n\nTop 1% in Wheat and Melanoma..............about to take down medals if Kaggle does not do something about it.", "votes": 2 }, { "id": 974950, "postDate": "2020-08-18T05:09:40.153Z", "content": "Some of them does have skills. Pity to see the way using it…
", "rawMarkdown": "Some of them does have skills. Pity to see the way using it..." } ] }, { "id": 973246, "postDate": "2020-08-17T07:37:07.483Z", "content": "Nothing suspicious at all…
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F443651%2F5b26bb36db86ce13caab6d66107b333e%2Fprivate_sharing1.png?generation=1597649824898942&alt=media\")
@group16 lol 4 submissions. All of those team mates are in on it too, these aren't random team meetups. It seems to have been going on for at least 2 years, but probably more based on what I saw on the website before @lifeifei38 took it down.
", "rawMarkdown": "@group16 lol 4 submissions. All of those team mates are in on it too, these aren't random team meetups. It seems to have been going on for at least 2 years, but probably more based on what I saw on the website before @lifeifei38 took it down.", "votes": 3 }, { "id": 973258, "postDate": "2020-08-17T07:47:22.577Z", "content": "Exactly. The low submission count & the team of novices is what made them stood out in the first place. Both of them have a abnormal low amount of submissions to score that high, which indicates that other teams were involved as well.
\nIt is weird that this team is so successful, and that 4-5 of the team members haven't even bothered to get the contributor tier (which would require them to verify themselves).
", "rawMarkdown": "Exactly. The low submission count & the team of novices is what made them stood out in the first place. Both of them have a abnormal low amount of submissions to score that high, which indicates that other teams were involved as well.\n\nIt is weird that this team is so successful, and that 4-5 of the team members haven't even bothered to get the contributor tier (which would require them to verify themselves).", "votes": 2 }, { "id": 973287, "postDate": "2020-08-17T08:03:24.513Z", "content": "@group16 so they must not be aspiring Data Scientists. They must just be mules, farm accounts, and by not filling in any bio or info, that probably makes the account more valuable, as its shiny and new. How is it, a few of us pull this info together in minutes and nothing gets done about these people? They are seriously going to take medals, possibly gold, in both Wheat and Melanoma in the next 15 hours. I'm really raging about this, i seriously wish we had access to data to build a node graph with edges, just throw in the dozen or two names we have already doxed and see all who is fruit of the poisonous tree.
", "rawMarkdown": "@group16 so they must not be aspiring Data Scientists. They must just be mules, farm accounts, and by not filling in any bio or info, that probably makes the account more valuable, as its shiny and new. How is it, a few of us pull this info together in minutes and nothing gets done about these people? They are seriously going to take medals, possibly gold, in both Wheat and Melanoma in the next 15 hours. I'm really raging about this, i seriously wish we had access to data to build a node graph with edges, just throw in the dozen or two names we have already doxed and see all who is fruit of the poisonous tree.\n\n", "votes": 1 }, { "id": 973290, "postDate": "2020-08-17T08:06:54.453Z", "content": "It should be possible to do graph analysis based on the Kaggle Meta-dataset. It's going to be quite a bit of work though, but it would indeed be interested to see. Pretty sure we can create a large graph that connects all of these accounts together.
", "rawMarkdown": "It should be possible to do graph analysis based on the Kaggle Meta-dataset. It's going to be quite a bit of work though, but it would indeed be interested to see. Pretty sure we can create a large graph that connects all of these accounts together." }, { "id": 973386, "postDate": "2020-08-17T09:22:13.010Z", "content": "Sounds like a plan
", "rawMarkdown": "Sounds like a plan" } ] } ] }, { "id": 970355, "postDate": "2020-08-14T11:34:33.130Z", "content": "Kaggle should make sure to have some detection systems in place. I found something very suspicious during the Liverpool competition as well. I never checked up on it, but appears one of them have gotten silver medals (the other probably used multiple accounts and got banned for that). Both of them are now in a team together in the wheat detection competition…
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F443651%2F35281b0641e4c33537c8321370716d45%2Fliverpool_private_sharing.png?generation=1597404868770947&alt=media\")
There isn't any doubt they cheated. Both accounts should be banned from Kaggle.
", "rawMarkdown": "There isn't any doubt they cheated. Both accounts should be banned from Kaggle.", "votes": 4 }, { "id": 970412, "postDate": "2020-08-14T12:46:18.643Z", "content": "![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2Fc84856f4cdcc46066b17b590d2059e75%2F_20200814143921.png?generation=1597409127814306&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2Fc60074c3789d2405d015ed0acda8b6d9%2F_20200814143954.png?generation=1597409147910064&alt=media\")
Enjoy!
", "rawMarkdown": "\n\n\n\nEnjoy!", "votes": 16 }, { "id": 973178, "postDate": "2020-08-17T06:43:43.520Z", "content": "This is so funny and sad 😄
", "rawMarkdown": "This is so funny and sad 😄" } ] }, { "id": 972892, "postDate": "2020-08-17T00:20:28.593Z", "content": "I guess this can be related to https://edu.angtk.com/course/35 ? This competition worths $1000?
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1343903%2Feb3a71ae5dd1b9304fb9f568b83c636c%2Fa.png?generation=1597624521014125&alt=media\")
Edit: the website is down now.
", "rawMarkdown": "I guess this can be related to https://edu.angtk.com/course/35 ? This competition worths $1000?\n\n\nEdit: the website is down now. \n", "votes": 14, "replies": [ { "id": 972906, "postDate": "2020-08-17T00:59:43.440Z", "content": "That website mentions some people who \"teach\" by name:
\nZhang Peng
\nGuo Junlin
\nChen Zidong
\nLi Xiaoyu
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F69253%2Fc641483a50a3a578f220e0c7bee35d27%2FScreen%20Shot%202020-08-16%20at%209.05.16%20PM.png?generation=1597626610346732&alt=media\")
I assume these people are on Kaggle?
", "rawMarkdown": "That website mentions some people who \"teach\" by name:\n\nZhang Peng\nGuo Junlin\nChen Zidong\nLi Xiaoyu\n\n\n\nI assume these people are on Kaggle?", "votes": 9 }, { "id": 972926, "postDate": "2020-08-17T01:27:58.580Z", "rawMarkdown": "", "isDeleted": true }, { "id": 972943, "postDate": "2020-08-17T01:43:44.633Z", "rawMarkdown": "", "votes": 4, "isDeleted": true }, { "id": 972958, "postDate": "2020-08-17T02:02:29.273Z", "rawMarkdown": "", "votes": 3, "isDeleted": true }, { "id": 972965, "postDate": "2020-08-17T02:15:46.937Z", "content": "We are just a teaching website. All teaching materials can be provided to the organizers for viewing. All of them are open baseline procedures. And we don't offer to know what to share in private. This video, we can put on the youtube video website, just explained some basic knowledge of public baseline!
", "rawMarkdown": "We are just a teaching website. All teaching materials can be provided to the organizers for viewing. All of them are open baseline procedures. And we don't offer to know what to share in private. This video, we can put on the youtube video website, just explained some basic knowledge of public baseline!", "votes": -19 }, { "id": 972969, "postDate": "2020-08-17T02:20:32.367Z", "content": "I can be sure that our students have participated in this competition. They all use the open baseline to learn. But not in the gold medal area. All methods are based on the discussion group here or in notebooks.
", "rawMarkdown": "I can be sure that our students have participated in this competition. They all use the open baseline to learn. But not in the gold medal area. All methods are based on the discussion group here or in notebooks.", "votes": -17, "replies": [ { "id": 973012, "postDate": "2020-08-17T03:28:25.350Z", "content": "And they pay $1000? to people like you who have never even posted on Kaggle until today? Never entered a comp? Do all the instructors have accounts with 0 competitions entered and no posts in the Kaggle community? I mean, you realize how this looks right? You just identified yourself as one of the instructors. Yet you also have an account @fffeiyee which has entered comps with other \"instructors\" on that site such as @dongxiaoming . You realize having multiple accounts is against the rules right?
", "rawMarkdown": "And they pay $1000? to people like you who have never even posted on Kaggle until today? Never entered a comp? Do all the instructors have accounts with 0 competitions entered and no posts in the Kaggle community? I mean, you realize how this looks right? You just identified yourself as one of the instructors. Yet you also have an account @fffeiyee which has entered comps with other \"instructors\" on that site such as @dongxiaoming . You realize having multiple accounts is against the rules right?", "votes": 16 }, { "id": 973026, "postDate": "2020-08-17T03:51:21.050Z", "content": "Under \"Student News\" for the Kaggle Melanoma class (offered at $1000), are two students……….and there names just happen to be……….wait for it………. Wang and Yang:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F69253%2Fc8c48f0006c67c71f9616c87ee68ac4b%2FScreen%20Shot%202020-08-16%20at%2011.49.21%20PM.png?generation=1597636279665678&alt=media\")
Thank you for your question. Let me answer it
\nFirst of all, I will clarify our knowledge website teaching and Q & A. The teacher is part-time! The screenshot is the video content of this website, and the video I explained is the video for basic teaching. We have no right to disclose the teacher's account number. They just record some basic teaching videos on a part-time basis. We have not organized students to participate in the competition. They will ask questions if they have questions, but they will not provide us with their accounts.
\nAnd the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this.
", "rawMarkdown": "Thank you for your question. Let me answer it\n\nFirst of all, I will clarify our knowledge website teaching and Q & A. The teacher is part-time! The screenshot is the video content of this website, and the video I explained is the video for basic teaching. We have no right to disclose the teacher's account number. They just record some basic teaching videos on a part-time basis. We have not organized students to participate in the competition. They will ask questions if they have questions, but they will not provide us with their accounts.\n\nAnd the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this.\n\n", "votes": -7 }, { "id": 973033, "postDate": "2020-08-17T04:00:59.853Z", "content": "Watching the same basic video doesn't mean sharing it in private. This video is not related to match scoring. Just use the baseline to explain the basis of deep learning. I said, I will release these videos tomorrow and then youtube, we are registering and uploading!
", "rawMarkdown": "Watching the same basic video doesn't mean sharing it in private. This video is not related to match scoring. Just use the baseline to explain the basis of deep learning. I said, I will release these videos tomorrow and then youtube, we are registering and uploading!", "votes": -11 }, { "id": 973054, "postDate": "2020-08-17T04:41:41.810Z", "content": "\n\nAnd the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this.
\n
How much does it cost? Even though it doesn't cost that much, it looks fishy.
", "rawMarkdown": "> And the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this.\n\nHow much does it cost? Even though it doesn't cost that much, it looks fishy.", "votes": 2 }, { "id": 973203, "postDate": "2020-08-17T07:02:50.917Z", "content": "@lifeifei38 Why did you just take the entire web site down? You must be the/one of the kingpins. Making money off selling Kaggle comps and you don't even participate. No doubt lifefei38 is a fake account, it has no activity, accept this thread………I hope your real identity is discovered.
", "rawMarkdown": "@lifeifei38 Why did you just take the entire web site down? You must be the/one of the kingpins. Making money off selling Kaggle comps and you don't even participate. No doubt lifefei38 is a fake account, it has no activity, accept this thread.........I hope your real identity is discovered. ", "votes": 3 }, { "id": 973236, "postDate": "2020-08-17T07:28:32.993Z", "content": "We are really sorry that we used an incorrect word in the sentence \"Our actual cost is lower than this.\" We wanted to say \"The actual price of our lesson is lower than the price showing in our website.\" This is a language problem, we carelessly used the word \"cost\" by mistake. The lessons and their prices on our website is just for show, while when a student want to attend our lesson, he/she contact us by instant messaging application and we will further discuss the price concerning his/her needs and open the suitable lesson(s) for him/her.
\nWe will upload lessons to Youtube tomorrow. The lessons are in Chinese but we will use automatic or human translation to add English subtitles and let you (also Kaggle and other Kagglers) know that our lessons
\ninclude only basic explanations, but not the hints for improving the score of the competition.
\nA lot of cross-disciplinary beginners for Kaggle have difficulties taking part in the competitions because they are unfamiliar with the operation of Kaggle or have questions concerning the basic knowledge, this is where our lessons helps.
I just explain the video content of our website.There are a lot of similar videos. On youtube, you can go there and grab account. We have an obligation to protect our web account information
", "rawMarkdown": "I just explain the video content of our website.There are a lot of similar videos. On youtube, you can go there and grab account. We have an obligation to protect our web account information", "votes": -11 } ] }, { "id": 972975, "postDate": "2020-08-17T02:27:34.333Z", "content": "We can put all three videos of this course on YouTube tomorrow, August 18. And this video hasn't been updated half a month ago. We only provide the most basic baseline learning lectures, not the sharing of competition skills.
", "rawMarkdown": "We can put all three videos of this course on YouTube tomorrow, August 18. And this video hasn't been updated half a month ago. We only provide the most basic baseline learning lectures, not the sharing of competition skills.", "votes": -2 }, { "id": 973008, "postDate": "2020-08-17T03:14:08.353Z", "content": "Yeah, you don't offer to know. We know something called social media with digital pay in it.
\nThanks for your explanation to clear our doubts. Your four videos each len 40 minutes worth 1000dollars. There must be sth really serious and hard stuff for us to understand. So for more dicretion and easy understanding, why don't u guys just share your teachers' and students' real kaggle profile page and email for the organizer to investigate. We see that you have many teachers and 36 students, right? They must be on kaggle and submitted baselines, right? So they can be evaluated? And your teachers must be very experienced and good on kaggle so people pay 1000 dollars to listen for their less than 3 hours speech, right? GrandMaster maybe?
\nI'm very happy to see those genius novices on kaggle who will in the future do great job for AI development and benefit more people with their amazing skills. Those people are hopes for AI's next stage fast development.
\nThanks again for your work in teaching and your help to thrive the kaggle community by teaching others.
", "rawMarkdown": "Yeah, you don't offer to know. We know something called social media with digital pay in it.\n\nThanks for your explanation to clear our doubts. Your four videos each len 40 minutes worth 1000dollars. There must be sth really serious and hard stuff for us to understand. So for more dicretion and easy understanding, why don't u guys just share your teachers' and students' real kaggle profile page and email for the organizer to investigate. We see that you have many teachers and 36 students, right? They must be on kaggle and submitted baselines, right? So they can be evaluated? And your teachers must be very experienced and good on kaggle so people pay 1000 dollars to listen for their less than 3 hours speech, right? GrandMaster maybe?\n\nI'm very happy to see those genius novices on kaggle who will in the future do great job for AI development and benefit more people with their amazing skills. Those people are hopes for AI's next stage fast development. \n\nThanks again for your work in teaching and your help to thrive the kaggle community by teaching others.", "votes": 14 }, { "id": 973020, "postDate": "2020-08-17T03:43:59.523Z", "content": "First of all, thank you very much for your reasonably requesting us to clarify the whole matter.
\nOur students just come to study, I don't know their account number. In addition, we have online basic Q & A, as well as basic questions of deep learning and machine learning on this website. And the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this. Many teachers are only responsible for explaining the basic knowledge concerning the baseline and answering questions. In the future, we will organize teachers to share some notebooks, but now we have no plan to disclose personal information as we need to ask for personal consent. In addition, we can not control whether the students will discuss in other places after watching the course. We are just an online teaching and answering website, not an actual university with regulations for students.
", "rawMarkdown": "First of all, thank you very much for your reasonably requesting us to clarify the whole matter.\n\nOur students just come to study, I don't know their account number. In addition, we have online basic Q & A, as well as basic questions of deep learning and machine learning on this website. And the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this. Many teachers are only responsible for explaining the basic knowledge concerning the baseline and answering questions. In the future, we will organize teachers to share some notebooks, but now we have no plan to disclose personal information as we need to ask for personal consent. In addition, we can not control whether the students will discuss in other places after watching the course. We are just an online teaching and answering website, not an actual university with regulations for students.", "votes": -4 }, { "id": 973063, "postDate": "2020-08-17T04:55:19.057Z", "rawMarkdown": "", "isDeleted": true }, { "id": 973066, "postDate": "2020-08-17T04:59:13.750Z", "rawMarkdown": "", "isDeleted": true }, { "id": 973096, "postDate": "2020-08-17T05:27:41.373Z", "content": "You know you just said that you guys really take money from this. Where did those happen, where did your guys get paid since it't not on the official webiste? How many of those secret payments relating to private \"baseline explanaion\" & \"Q&A\" etc happened? We don't know; The kaggle officials don't know; You may or may not know.
\nSeriously, is this account your offical account on kaggle? Should people focus this account if he/she wants to learn in your \"community\"?
\nDude, this thread is gonna be very famous and controversial. A really famous one thanks to you from an account only in charge of discussion under this thread with thousands of words revealing what you have done. And yet you dodge key questions- real kaggle profiles of your precious teachers must also be private because they are that kind of online teachers serving a certain group of \"student\" who paid well and don't like others to see them? Right, a bit strange for teachers on a ML&DL teaching webisite maybe? Cause teachers elsewhere online just want more people to know them.
\nI think I'm not gonna to put any more time into this shit. Keep talking and let more people to know this thread, to know what your guys are and what you've done and your clarification. Good job and keep working! Maybe I'll see this thread in next BigDataDigest news - finally someone behind those genius novices come up, a new novice account with zero competition experience but not GrandMaster could be disappointed. But still, Great source! Great ad! Good for your business!
", "rawMarkdown": "You know you just said that you guys really take money from this. Where did those happen, where did your guys get paid since it't not on the official webiste? How many of those secret payments relating to private \"baseline explanaion\" & \"Q&A\" etc happened? We don't know; The kaggle officials don't know; You may or may not know.\n\nSeriously, is this account your offical account on kaggle? Should people focus this account if he/she wants to learn in your \"community\"?\n\nDude, this thread is gonna be very famous and controversial. A really famous one thanks to you from an account only in charge of discussion under this thread with thousands of words revealing what you have done. And yet you dodge key questions- real kaggle profiles of your precious teachers must also be private because they are that kind of online teachers serving a certain group of \"student\" who paid well and don't like others to see them? Right, a bit strange for teachers on a ML&DL teaching webisite maybe? Cause teachers elsewhere online just want more people to know them.\n\nI think I'm not gonna to put any more time into this shit. Keep talking and let more people to know this thread, to know what your guys are and what you've done and your clarification. Good job and keep working! Maybe I'll see this thread in next BigDataDigest news - finally someone behind those genius novices come up, a new novice account with zero competition experience but not GrandMaster could be disappointed. But still, Great source! Great ad! Good for your business! ", "votes": 8, "replies": [ { "id": 973138, "postDate": "2020-08-17T06:07:17.540Z", "content": "Want to be sick to your stomach? I just went to their M5 \"class\" page, and it shows \"New Students\". It shows this for each class. Each class is full of brand new kaggle accounts, with no activity, no post history, just enter the competition, with other accounts who also are new and have no history. and get high scores. Here is just what I manually looked up on M5. Also realize each of these is teamed with other people, who also are newbies, and those people team with people, and it goes on and on…….
\nScraped from Link for M5 competition “class” “Newest Students” at this link: https://translate.google.com/translate?hl=en&sl=zh-CN&u=https://edu.angtk.com/course/32/notes%3Fsort%3Dlike%26task%3D465%26selectedCourse%3D0&prev=search&pto=aue
\nStudent 1
\naustin4052 aka Austin Kang, Appears in Kaggle for the first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 559 out of 5500+ people just barely missing a bronze.
\nTeamed with
\n 红警free ( liyishuolys)
\n Zhengbo ZHOU (curryzhou)
\n Ningcan Sun (ningcansun)
\n Jianzhong Tian (jianzhongtian)
Student 2
\nEric_Tonser Appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 551 out of 5500+ people medaling a bronze.
\nTeamed with
\n 不正常人类研究中心 (sjbowen)
\n 小鱼仔 (xiaoyuzai)
\n Cherish_L(cherishl)
\n KeepingFly (keepingfly)
Student 3
\nJesse (Kaggle info unknown)
Student 4
\nSun Ningcan appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 559 out of 5500+ people just barely missing a bronze.
\nTeamed with
\n 红警free ( liyishuolys)
\n Zhengbo ZHOU (curryzhou)
\n Austin Kang
\n Jianzhong Tian (jianzhongtian)
Student 5
\nHxysophy (Kaggle info unknown)
Student 6
\nKoko (Kaggle info unknown)
Student 7
\nMadokazu (Kaggle info unknown)
Student 8
\nKikkk which looks awfully similar to kikikkk which appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 1325 out of 5500+
\nTeamed with:
\n Henry Tang (henryshtang)
\n Mengton Xu (mengtongxu)
\n Xiang Anqi (xianganqi)
Student 9
\nSinner0612 (Kaggle info unknown)
Student 10
\nericyuu (Kaggle info unknown)
Student 11
\nSnowmeli (snowmeli) appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 490 out of 5500+
\nTeamed with
\n машинное обучение (a17186819035)
\n Hubert He (anuhw0228)
\n WeiHao Li (josephliweihao)
Student 12
\nanuhw (Kaggle info unknown)
Student 13
\nliweihao (liweihao) interesting that it is the same “name” as the person I just mentioned above WeiHao Li (josephliweihao)
Student 14
\nTaohuahuahuanyugan (Kaggle info unknown)
Student 15
\nWoww (Kaggle info unknown)
Student 16
\nJRiY (Kaggle info unknown)
Student 17
\nPowdererer (Kaggle info unknown)
Student 18
\nquanbingtao (Kaggle info unknown)
Student 19
\nchiaraduan (Kaggle info unknown)
Student 20
\nMaster Yi Star
Google's history page clearly indicates that what we do is EDA and simple baseline submission. It doesn't involve any scoring skills and tricks. Is it against the rules here to submit several notebooks and not get medals? In addition, it is against the law to disclose information that we don't want to control others?![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F5466487%2Ff93c9da0fac0e786ac313ba246ce27f4%2F2020-08-17%20142509.png?generation=1597645661210321&alt=media\")
That's definitely against the rules. It's called private sharing. I remember things being shared on a slack channel during the instant gratification competition, which was also classified as private sharing.
", "rawMarkdown": "That's definitely against the rules. It's called private sharing. I remember things being shared on a slack channel during the instant gratification competition, which was also classified as private sharing.", "votes": 16 }, { "id": 973167, "postDate": "2020-08-17T06:34:51.037Z", "content": "If you did nothing against the rules, you can share the kaggle profiles of your teachers and students.
", "rawMarkdown": "If you did nothing against the rules, you can share the kaggle profiles of your teachers and students. ", "votes": 6 }, { "id": 973172, "postDate": "2020-08-17T06:38:40.203Z", "content": "The students just came here to study, and they didn't tell us what the account number is. We only answer questions, such as their registration or notebook questions. And we can also see that we use the open source baseline of kaggle to do some more detailed technical explanation.
", "rawMarkdown": "The students just came here to study, and they didn't tell us what the account number is. We only answer questions, such as their registration or notebook questions. And we can also see that we use the open source baseline of kaggle to do some more detailed technical explanation.", "votes": -5 }, { "id": 973177, "postDate": "2020-08-17T06:43:23.997Z", "rawMarkdown": "", "isDeleted": true }, { "id": 973179, "postDate": "2020-08-17T06:43:57.340Z", "content": "Amazing! Here are some of your students in Melanoma.
\nWhats fascinating is that all your students join Kaggle. Never post, never comment, have no activity. They join large teams. Everyone in the team is the same, they are all new, no activity. The teams all score near the top, some medal.
\n\nStudent 1
\nGawain Gan (gawaingan) Top 11% M5 Forecasting, Top 12% in Melanoma, Top 26% in Prostate
\nTeamed with (Melanoma)
\n Watmics (watmics)
\n Jiaqing Xie (godxie)
\n Sifan Wang96 (sifanwang96)
\n gw.w9 (guoweiw9)
Student 2
\nBochen (Kaggle info unknown)
Student 3
\nmath
Student 4
\nZhao Yujie (zhaoyujie123) Currently 209 in Melanoma. Like everyone in this team has done the same thing, new account, no posts, no comments, etc, etc.
\nTeamed with
\n Patrick (tangyaxu)
\n Du Shaohui (dushaohui)
\n Xinjian (xinjianli)
\n Noone (neyiii)
Student 5
\nJoe1022 (Kaggle info unknown)
Student 6
\nTangyaxu Patrick (tangyaxu). Mentioned in above team
Student 7
\nGw_wang (Kaggle info unknown)
Student 8
\nMadokazu (Kaggle info unknown) Mentioned in M5 list too
Student 9
\nSnowmeli (snowmeli) was in M5 and prostrate
Student 10
\nAnuhw (Kaggle info unknown)
Student 11
\nLiweihao (liweihao)
Student 12
\nChangbc777 Siyu Guan (changbc777) no comps entered yet, no activity
Student 13
\nJIAQINGXIE (godxie) mentioned above in team, and set to medal
Student 14
\nHobart01 (Kaggle info unknown)
Student 15
\nzzzjh (Kaggle info unknown)
Student 16
\nLiweihao (liweihao) close to someone in my other list , same “name” as the person I just mentioned above WeiHao Li (josephliweihao)
Student 17
\nXmetric (Kaggle info unknown)
Student 18
\nsmrqqqq (Kaggle info unknown)
Student 19
\nJasonHuang (Kaggle info unknown)
Student 20
\njiacheng_shi (Kaggle info unknown)
Student 21
\naskpatrickgor (Kaggle info unknown)
Sharing notebooks only with people that pay for it and making them not publicly available and known to everyone is the definition of private sharing.
", "rawMarkdown": "Sharing notebooks only with people that pay for it and making them not publicly available and known to everyone is the definition of private sharing.", "votes": 6 }, { "id": 973207, "postDate": "2020-08-17T07:07:21.840Z", "content": "What we use to teach is an public baseline in notebook.
", "rawMarkdown": "What we use to teach is an public baseline in notebook.", "votes": -1 }, { "id": 973326, "postDate": "2020-08-17T08:45:11.717Z", "content": "You keep saying that you are teaching them using public kernels , what's your addon ? why would someone pay for public information they can go on kaggle and see public notebooks and if they have any questions I'm sure the author will answer them, my opinion is they're paying to get push on lb whether it's a submission file or leaks that you are probably not allowed to share, if not why would they pay you ? ….
", "rawMarkdown": "You keep saying that you are teaching them using public kernels , what's your addon ? why would someone pay for public information they can go on kaggle and see public notebooks and if they have any questions I'm sure the author will answer them, my opinion is they're paying to get push on lb whether it's a submission file or leaks that you are probably not allowed to share, if not why would they pay you ? ....", "votes": 9 } ] }, { "id": 973252, "postDate": "2020-08-17T07:40:09.493Z", "rawMarkdown": "", "votes": 3, "isDeleted": true }, { "id": 973275, "postDate": "2020-08-17T07:56:22.480Z", "content": "I took many screenshots before @lifeifei38 took it down. I was using translate but many of the writings are pics not text.
\nHere is a dropbox with a few of the pics:
\nhttps://www.dropbox.com/sh/mev70eeh7tvln0g/AAAyGMupYhMTZ3lo1FIeZc7Ha?dl=0
It is very frustrating to see my position going down on LB so fast…
\nI am a beginner and this really is disappointing to see.
\nI am not against public kernels as long as they just give ideas ( some great public kernels have helped me throughout this competition…increasing my knowledge ), but when they directly give the submission file…it is very unfair..
Thank you everyone to be active on discussions and publishing kernels with new ideas.
", "rawMarkdown": "It is very frustrating to see my position going down on LB so fast...\nI am a beginner and this really is disappointing to see.\nI am not against public kernels as long as they just give ideas ( some great public kernels have helped me throughout this competition...increasing my knowledge ), but when they directly give the submission file...it is very unfair..\n\nThank you everyone to be active on discussions and publishing kernels with new ideas.", "votes": 11, "replies": [ { "id": 979138, "postDate": "2020-08-20T16:42:24.077Z", "content": "why you are not showing up on LB ? I have seen you when competition was going on.
\nwhat was your rank ?
Is it possible to remove yourself from a comp once you have joined? I don't think so.
", "rawMarkdown": "Is it possible to remove yourself from a comp once you have joined? I don't think so." } ] }, { "id": 970572, "postDate": "2020-08-14T15:19:01.517Z", "content": "We need Kaggle competition \"Find fake account in Kaggle\" and it is important to find real main account to punish cheater (remove medals or bans or something else). In every online game we have cheaters, so in Kaggle, and it is always annoying
", "rawMarkdown": "We need Kaggle competition \"Find fake account in Kaggle\" and it is important to find real main account to punish cheater (remove medals or bans or something else). In every online game we have cheaters, so in Kaggle, and it is always annoying", "votes": 12, "replies": [ { "id": 970582, "postDate": "2020-08-14T15:24:08.087Z", "content": "I think Kaggle's algorithm to remove cheaters really needs to be improved
", "rawMarkdown": "I think Kaggle's algorithm to remove cheaters really needs to be improved", "votes": 13 }, { "id": 974258, "postDate": "2020-08-17T20:44:51.420Z", "content": "Having said that, where is Kaggle in this thread?? …Never mind, they have arrived.
", "rawMarkdown": "Having said that, where is Kaggle in this thread?? ...Never mind, they have arrived.", "replies": [ { "id": 974264, "postDate": "2020-08-17T20:53:35.140Z", "content": "The silence isn't comforting, but at the same time, this is a rather large can of worms that just got opened so I am sure it's quite a bit for them to deal with.
", "rawMarkdown": "The silence isn't comforting, but at the same time, this is a rather large can of worms that just got opened so I am sure it's quite a bit for them to deal with. " } ] } ] }, { "id": 970404, "postDate": "2020-08-14T12:32:37.303Z", "content": "From liverpool to Trends to Alaska to Wheat detection to this one to even landmark 2020. Kaggle has done little effort, even now some of those competitions have been finalized, those LB still contain many cheaters.
", "rawMarkdown": "From liverpool to Trends to Alaska to Wheat detection to this one to even landmark 2020. Kaggle has done little effort, even now some of those competitions have been finalized, those LB still contain many cheaters.", "votes": 12, "replies": [ { "id": 973297, "postDate": "2020-08-17T08:17:15.330Z", "content": "yikououbao has taken 3 medals so far, about to take 2 more…..it's not just one team though. Some are more obvious than others, and you can certainly cross-reference the team members with other teams they have been in to uncover more of the conspirators, but it comes down to that there just aren't the right controls in place to prevent this sort of thing.
", "rawMarkdown": "yikououbao has taken 3 medals so far, about to take 2 more.....it's not just one team though. Some are more obvious than others, and you can certainly cross-reference the team members with other teams they have been in to uncover more of the conspirators, but it comes down to that there just aren't the right controls in place to prevent this sort of thing.", "votes": 2 } ] }, { "id": 973654, "postDate": "2020-08-17T12:38:03.653Z", "content": "One thing I learned is that loading all of Users.csv and TeamMembership.csv from Kaggles Meta Dataset makes Gephi drop to its knees, lol.
If someone has some time an interesting Node Graph would be:
\nUsers.csv)TeamMemberships.csv)TeamMemberships.csv)Additionally, you would want to join the User id's you find in Step 2, using Users.csv as Labels. Same with Teams instead of just Team ID's you could join the ID's in TeamMemberships.csv with Team labels found in Teams.csv.
I think you will find that its a closed loop. The networks will only grow so far. All will contain the same types of users. Users with no bio, no comments, no posts, immediately entering competitions and getting high scores.
\nGood tools for doing this are Gephi (slow), NetworkX, Neo4j
\nGreat tools for doing this would be cuGraph (RAPIDS). I will do some myself, but I have not slept in over 24 hours, and I know time is not our friend when medals stand to be lost in a matter of hours.
A list of the Users mentioned in this thread is below:
\n\"Id\",\"UserName\",\"DisplayName\",\"RegisterDate\",\"PerformanceTier\"\n\"2607029\",\"dlwangyp\",\"DLWangYP\",\"12/12/2018\",\"0\"\n\"1834362\",\"chenggj\",\"cheng\",\"04/18/2018\",\"0\"\n\"5214874\",\"xueyang0827\",\"Xue Yang 0827\",\"06/02/2020\",\"0\"\n\"5581487\",\"yosefhuang\",\"yosef huang\",\"08/06/2020\",\"0\"\n\"2301868\",\"lg920810\",\"幼稚园接悟空\",\"09/29/2018\",\"0\"\n\"5259643\",\"birdfly123\",\"birdfly123\",\"06/08/2020\",\"0\"\n\"5582725\",\"zhaogongzuo2\",\"zhaogongzuo2\",\"08/06/2020\",\"0\"\n\"5413611\",\"marwafatto\",\"Marwa_Fatto\",\"07/04/2020\",\"0\"\n\"2592536\",\"zengchan\",\"zengchan\",\"12/08/2018\",\"0\"\n\"3859401\",\"silewang\",\"480073534\",\"10/15/2019\",\"0\"\n\"4198084\",\"yikououbao\",\"yikououbao\",\"12/13/2019\",\"2\"\n\"4360898\",\"eliowang\",\"Edward\",\"01/21/2020\",\"0\"\n\"5285253\",\"yauwaiyim\",\"Yau Wai Yim\",\"06/12/2020\",\"0\"\n\"5366857\",\"siqicao\",\"siqi cao\",\"06/25/2020\",\"0\"\n\"2000407\",\"austin4052\",\"Austin Kang\",\"06/17/2018\",\"0\"\n\"3101840\",\"liyishuolys\",\"红警free\",\"04/17/2019\",\"0\"\n\"3858729\",\"curryzhou\",\"Zhengbo ZHOU\",\"10/15/2019\",\"0\"\n\"5292127\",\"ningcansun\",\"Ningcan Sun\",\"06/13/2020\",\"0\"\n\"5346060\",\"jianzhongtian\",\"Jianzhong Tian\",\"06/22/2020\",\"0\"\n\"5321313\",\"tonserboyt\",\"TonserBoyT\",\"06/17/2020\",\"0\"\n\"2438897\",\"sjbowen\",\"不正常人类研究中心\",\"11/02/2018\",\"0\"\n\"4100810\",\"xiaoyuzai\",\"小鱼仔\",\"11/25/2019\",\"0\"\n\"4321861\",\"cherishl\",\"Cherish_L\",\"01/13/2020\",\"1\"\n\"5175992\",\"keepingfly\",\"KeepingFly\",\"05/27/2020\",\"0\"\n\"4672128\",\"kikikkk\",\"kikikkk\",\"03/15/2020\",\"0\"\n\"1072650\",\"henryshtang\",\"Henry Tang\",\"05/13/2017\",\"1\"\n\"4671848\",\"mengtongxu\",\"Mengtong XU\",\"03/15/2020\",\"0\"\n\"4672134\",\"xianganqi\",\"XIANG ANQI\",\"03/15/2020\",\"0\"\n\"5345886\",\"snowmeli\",\"Snowmeli\",\"06/22/2020\",\"0\"\n\"5129046\",\"a17186819035\",\"машинное обучение\",\"05/20/2020\",\"0\"\n\"5345891\",\"anuhw0228\",\"Hubert He\",\"06/22/2020\",\"1\"\n\"5345986\",\"josephliweihao\",\"WeiHao Li\",\"06/22/2020\",\"0\"\n\"2137547\",\"gawaingan\",\"Gawain Gan\",\"08/08/2018\",\"0\"\n\"2379406\",\"watmics\",\"Watmics\",\"10/18/2018\",\"1\"\n\"3648601\",\"godxie\",\"JIAQING XIE\",\"08/31/2019\",\"1\"\n\"4600383\",\"sifanwang96\",\"Sifan wang96\",\"03/03/2020\",\"0\"\n\"5067153\",\"guoweiw9\",\"gw.w9\",\"05/11/2020\",\"0\"\n\"4672655\",\"zhaoyujie123\",\"ZhaoYuJie 123\",\"03/15/2020\",\"0\"\n\"3422743\",\"tangyaxu\",\"Patrick\",\"07/03/2019\",\"1\"\n\"4600143\",\"dushaohui\",\"Du Shaohui\",\"03/03/2020\",\"0\"\n\"4711749\",\"xinjianli\",\"Xinjian\",\"03/21/2020\",\"1\"\n\"5202248\",\"oneyiii\",\"noone\",\"05/31/2020\",\"0\"\nHere's a simple/ugly code snippet to get an already ugly visualization in python with networkx:
\nimport numpy as np\nimport pandas as pd\nimport networkx as nx\nimport matplotlib.pyplot as plt\n\nusers = pd.read_csv('../input/meta-kaggle/Users.csv')\nteams = pd.read_csv('../input/meta-kaggle/Teams.csv')\nteammembers = pd.read_csv('../input/meta-kaggle/TeamMemberships.csv')\n\nROOT = <ENTER USER ID HERE>\nnodes = {ROOT}\ndone = set()\n\ngraph = nx.DiGraph()\nfor _ in range(DEPTH): # How far/hops do you want to go from this node?\n to_visit = list(nodes)\n for node in to_visit:\n print(node)\n done.add(node)\n\n teams = teammembers[teammembers['UserId'] == node]['TeamId'].values\n for team in teams:\n for user in teammembers[teammembers['TeamId'] == team]['UserId']:\n if user not in done:\n nodes.add(user)\n\n graph.add_edge(node, user)\n\n nodes.remove(node)\n\npos = nx.drawing.layout.random_layout(graph)\nplt.figure(figsize=(15, 15))\nnx.drawing.nx_pylab.draw(graph, pos=pos)\nnx.drawing.nx_pylab.draw_networkx_labels(graph, pos=pos)\nplt.show()\nAll those customers of Kaggle solutions should be banned FOREVER, shame on them.
", "rawMarkdown": "All those customers of Kaggle solutions should be banned FOREVER, shame on them. ", "votes": 9 }, { "id": 972293, "postDate": "2020-08-16T12:40:03.923Z", "content": "Kaggle medal market strikes again
", "rawMarkdown": "Kaggle medal market strikes again", "votes": 10 }, { "id": 971542, "postDate": "2020-08-15T16:46:13.797Z", "content": "I must say that, after learning about the medal farm concept below, if any of the 'Magnificent 7' accounts survive scrutiny by the kaggle team, and also come out of the shakeup having obtained medals, it would be interesting to see the associated user activity over the next few months, for example, changes in Display Name, participation in other competitions, etc. i.e. activity indicative of a change in account ownership.
", "rawMarkdown": "I must say that, after learning about the *medal farm* concept below, if any of the ***'Magnificent 7'*** accounts survive scrutiny by the kaggle team, and also come out of the shakeup having obtained medals, it would be interesting to see the associated user activity over the next few months, for example, changes in Display Name, participation in other competitions, *etc*. i.e. activity indicative of a change in account ownership.", "votes": 10, "replies": [ { "id": 971672, "postDate": "2020-08-15T19:43:12.317Z", "content": "![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4712534%2F58ce0bf94ac92680bc15de8c2679b361%2Fmeme1%20(1).jpeg?generation=1597520577769862&alt=media\")
Hi all! Thanks for the info. We try to catch these but require conclusive evidence. Please submit information to https://www.kaggle.com/compliance if there is an undetected offense we should investigate.
", "rawMarkdown": "Hi all! Thanks for the info. We try to catch these but require conclusive evidence. Please submit information to https://www.kaggle.com/compliance if there is an undetected offense we should investigate.", "votes": 10, "replies": [ { "id": 970560, "postDate": "2020-08-14T15:02:40.423Z", "content": "Sometimes it's (nearly) impossible to get evidence, just by adding a bit bias to a high scoring submission, and submit it using a totally different ip address, you'll never know it. But it doesn't mean that it doesn't exist, in this post all cheats are quite obvious, but some less obvious ones do exist in this competition
", "rawMarkdown": "Sometimes it's (nearly) impossible to get evidence, just by adding a bit bias to a high scoring submission, and submit it using a totally different ip address, you'll never know it. But it doesn't mean that it doesn't exist, in this post all cheats are quite obvious, but some less obvious ones do exist in this competition", "votes": 4 }, { "id": 971834, "postDate": "2020-08-16T01:08:51.773Z", "content": "maybe it's time to require real user data when creating Kaggle account…? this way different IP won't help you hiding
", "rawMarkdown": "maybe it's time to require real user data when creating Kaggle account...? this way different IP won't help you hiding", "votes": 2 }, { "id": 974143, "postDate": "2020-08-17T18:40:52.327Z", "content": "What kind of real user data would you suggest?
\nI think it would be very difficult to spot fake IDs given the large number of nationalities competing here.
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3810582%2Ffcdee7cffe0ce1c99db3e89d90abc942%2Fkaggle_cheating_melanoma_3.png?generation=1597717335863743&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3810582%2F1e4a1be5077da7d278a916504c3cc151%2Fkaggle_cheating_melanoma_2.png?generation=1597717325902779&alt=media\")
https://www.kaggle.com/jmayank23
\nhttps://www.kaggle.com/mayankjain23
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1614008%2F5d1a58537ec9fb676907da05ef54effb%2FScreenshot%20(15).png?generation=1597715959046035&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1614008%2Fd4cbc3d16a0dff97484aca01fa10339e%2FScreenshot%20(14).png?generation=1597716001782142&alt=media\")
Wow what a coincidence! Same username, same everything, one placing 8th and the other 13th!
\nhttps://www.kaggle.com/ajibeshmohapatra
\nhttps://www.kaggle.com/jubunmohapatra
And both set to place around the same rank in Wheat as well!
\nMaybe I should change my name, lol
", "rawMarkdown": "Wow what a coincidence! Same username, same everything, one placing 8th and the other 13th!\n\nhttps://www.kaggle.com/ajibeshmohapatra\nhttps://www.kaggle.com/jubunmohapatra\n\nAnd both set to place around the same rank in Wheat as well!\n\nMaybe I should change my name, lol", "votes": 3 }, { "id": 974638, "postDate": "2020-08-18T02:09:15.907Z", "content": "look at the dataset names😂
\n8 place created fold 1 global wheat
\n13 place created fold 2 global wheat
SAME TIME CREATED ACCOUNTS 2 months ago
\ntaken part in SAME competition
\njust looking at the dataset namefold 2 global wheaton 1 acount andfold 1 global wheatin second account
\nsame name too
This is a smart strategy, so when they delete the one account, he will still have the other 😉
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F69253%2F61631966e1a9c573fe640f762fdfabee%2FScreen%20Shot%202020-08-17%20at%2010.09.28%20PM.png?generation=1597716700272764&alt=media\")
name literally says “#2”
", "rawMarkdown": " name literally says “#2” " } ] } ] }, { "id": 972959, "postDate": "2020-08-17T02:04:22.523Z", "content": "The fourth guy in current public LB named 'zhaogongzuo2' literally means 'to find a job' in Chinese pingyin. Why not if you just need to suddenly submit a great score in the final week and then get yourself one?
\nCertain group of people are becoming fearless on kaggle.
", "rawMarkdown": "The fourth guy in current public LB named 'zhaogongzuo2' literally means 'to find a job' in Chinese pingyin. Why not if you just need to suddenly submit a great score in the final week and then get yourself one?\n\nCertain group of people are becoming fearless on kaggle.", "votes": 7, "replies": [ { "id": 973321, "postDate": "2020-08-17T08:41:33.713Z", "content": "🤣 They are crazy
", "rawMarkdown": "🤣 They are crazy" } ] }, { "id": 972008, "postDate": "2020-08-16T06:48:52.250Z", "content": "someone just got 0.9655 using one submission, this is frustrating my team and I have been working for over a month to get to 0.9649 and bam just like that a novice gets higher ranking than you.
", "rawMarkdown": "someone just got 0.9655 using one submission, this is frustrating my team and I have been working for over a month to get to 0.9649 and bam just like that a novice gets higher ranking than you.", "votes": 7, "replies": [ { "id": 972078, "postDate": "2020-08-16T08:01:08.920Z", "content": "I think someone shared a notebook that scores ~0.965 just recently. (which should not have happened since it's less than 7 days until private LB reveal)
", "rawMarkdown": "I think someone shared a notebook that scores ~0.965 just recently. (which should not have happened since it's less than 7 days until private LB reveal)", "votes": 2 }, { "id": 972080, "postDate": "2020-08-16T08:01:16.023Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 971888, "postDate": "2020-08-16T03:46:36.997Z", "content": "Kaggle will remove them since they are novices. They would be allowed to cheat if they were Kaggle masters and grandmasters.
\nhttps://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard
\nhttps://www.kaggle.com/c/imet-2019-fgvc6/discussion/94430
This is bigger than you think. I was manually pulling up accounts from that site as fast as I could, and many are \"sleeper\" accounts. In Kaggle and have not even entered any comps yet. It's a mess. All those people cheated in Whale, DeepFake, M5, Pancreatic, now this comp, and so many more. People that would have gotten a medal, or a better medal…….cheated, so someone could make money.
", "rawMarkdown": "This is bigger than you think. I was manually pulling up accounts from that site as fast as I could, and many are \"sleeper\" accounts. In Kaggle and have not even entered any comps yet. It's a mess. All those people cheated in Whale, DeepFake, M5, Pancreatic, now this comp, and so many more. People that would have gotten a medal, or a better medal.......cheated, so someone could make money.", "votes": 3 }, { "id": 973200, "postDate": "2020-08-17T06:59:32.623Z", "content": "Ofc it’s ValeriyBabushkin from ods.ai😄
", "rawMarkdown": "Ofc it’s ValeriyBabushkin from ods.ai😄", "votes": 7 }, { "id": 973244, "postDate": "2020-08-17T07:36:39.773Z", "content": "https://www.kaggle.com/c/landmark-recognition-2019/leaderboard
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1424766%2F7a31ed130b6e028a2baca61f3d66fac3%2FScreenshot%202020-08-17%20at%2009.35.59.png?generation=1597649791971180&alt=media\")
I think @lifeifei38 did us a huge favor explaining that what they do is sharing public notebooks that are available for their students but not to the rest of the competitors.
\nThis argument is more than enough for kaggle to take actions because we know that in this case, kaggle algorithm won't detect any fraud, but the teacher himself explained the situation.
\nLet's believe that it's just a normal course explaining some deep learning basics and the students are smart enough to make their way to the medal zone with their own work and effort, they still worked with their course's pipelines that are not available for the rest of the competitors.
\nThanks @group16 for your investigation :)
\n", "rawMarkdown": "I think @lifeifei38 did us a huge favor explaining that what they do is **sharing public notebooks** that are available for their students but **not to the rest of the competitors.**\n\nThis argument is more than enough for kaggle to take actions because we know that in this case, kaggle algorithm won't detect any fraud, but the teacher himself explained the situation.\n\nLet's believe that it's just a normal course explaining some deep learning basics and the students are smart enough to make their way to the medal zone with their own work and effort, they still worked with their course's pipelines that are not available for the rest of the competitors. \n\nThanks @group16 for your investigation :)\n\n@juliaelliott @philculliton", "votes": 5, "replies": [ { "id": 973538, "postDate": "2020-08-17T10:57:27.217Z", "content": "I'm highly skeptical that all those students 'are smart enough to make their way to the medal zone with their own work and effort'. Evidence suggest otherwise.
", "rawMarkdown": "I'm highly skeptical that all those students 'are smart enough to make their way to the medal zone with their own work and effort'. Evidence suggest otherwise." } ] }, { "id": 973481, "postDate": "2020-08-17T10:21:40.273Z", "content": "@group16 My personal respect man! Really great job.
", "rawMarkdown": "@group16 My personal respect man! Really great job. ", "votes": 5 }, { "id": 973265, "postDate": "2020-08-17T07:49:16.273Z", "content": "So scaring, they're like bots… Even now they continue to submit !
", "rawMarkdown": "So scaring, they're like bots... Even now they continue to submit !", "votes": 5 }, { "id": 973194, "postDate": "2020-08-17T06:55:08.753Z", "content": "This has been going on a LONG time.
\nThey just took the site down. Site had MANY kaggle comps, with students registered 20, 65, even over a hundred students in the Whale Tail comp.
\nStudents are all BRAND NEW kaggle accounts. No activity, no comments, no posts. They team with other students, typically 4-5 in a team. Everyone has the same deal, no comments, no activity, no posts. They ALWAYS place toward the top of the comp, either medaling or very close to it.
\nKaggle is infiltrated, this is a real mess.
", "rawMarkdown": "This has been going on a LONG time.\n\nThey just took the site down. Site had MANY kaggle comps, with students registered 20, 65, even over a hundred students in the Whale Tail comp. \n\nStudents are all BRAND NEW kaggle accounts. No activity, no comments, no posts. They team with other students, typically 4-5 in a team. Everyone has the same deal, no comments, no activity, no posts. They ALWAYS place toward the top of the comp, either medaling or very close to it.\n\nKaggle is infiltrated, this is a real mess.\n\n", "votes": 5 }, { "id": 973213, "postDate": "2020-08-17T07:10:22.270Z", "content": "so many rules broken or bent
", "rawMarkdown": "so many rules broken or bent", "votes": 5 }, { "id": 975009, "postDate": "2020-08-18T05:43:08.400Z", "content": "In the mind of a buyer of this kind of \"services\":
\nRecruiter: So,what you have done so far?
\nBought code/acc user : Gold medal on Kaggle in Public Leaderboard
\nRecruiter: You're hired.
\n==============================
\nRecruiter: So,what you have done so far?
\nBought code/acc user : Gold medal on Kaggle in Public Leaderboard.
\nRecruiter : We will let you know (Spoiler alert :They won't let you know, in most of the cases ).
\nJob candidate closes door
\nRecruiter : Surprise
", "rawMarkdown": "In the mind of a buyer of this kind of \"services\":\n\n\n## **Expectations**\n\n**Recruiter**: So,what you have done so far?\n\n**Bought code/acc user** : Gold medal on Kaggle in Public Leaderboard\n\n**Recruiter**: You're hired.\n\n==============================\n\n## **Reality** \n\n**Recruiter**: So,what you have done so far?\n\n**Bought code/acc user** : Gold medal on Kaggle in Public Leaderboard.\n\n####**Cross Validation** : **Hold my beer**🍻\n\n#### **Gilles** : **[I am about to end this man's whole career](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175156).**\n\n**Recruiter** : We will let you know (**Spoiler alert** :They won't let you know, in most of the cases ).\n\n****Job candidate closes door****\n\n**Recruiter** : [Surprise](https://youtu.be/Fkk9DI-8el4)", "votes": 3 }, { "id": 974154, "postDate": "2020-08-17T18:52:10.453Z", "content": "There are so many fake accounts holding high ranks on public LB.
\nThis team of 5 members with 4 novices is cheating too. All 4 of the novices didn't even make a single submission.
\n ![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3530433%2Fff35b19f4ba9ca4783d7150e4c823cd6%2FScreenshot%20from%202020-08-18%20001801.png?generation=1597690117324631&alt=media\")
\n@group16 you are right they are surely going to sell these accounts because I don't find any good reason behind doing this.
\nNone of the accounts were active in any past competitions. They are only logged in for participating in team.
\n[P.S. There are many such accounts.]
\n@juliaelliott @kaggle do something. They are stealing medals from deserved ones.
Please rest assured, we will be taking action.
", "rawMarkdown": "Please rest assured, we will be taking action.", "votes": 26 }, { "id": 974176, "postDate": "2020-08-17T19:12:31.580Z", "content": "Thanks…!
", "rawMarkdown": "Thanks...!", "votes": 1 }, { "id": 974191, "postDate": "2020-08-17T19:29:55.693Z", "content": "@inversion thanks for the confirmation.
", "rawMarkdown": "@inversion thanks for the confirmation. ", "votes": 1 }, { "id": 974301, "postDate": "2020-08-17T21:21:33.973Z", "content": "@prateek0x the yikououbao team is particularly brazen, they have taken many medals and stand to do so in two more comps in the next few hours. They have no shame.
", "rawMarkdown": "@prateek0x the yikououbao team is particularly brazen, they have taken many medals and stand to do so in two more comps in the next few hours. They have no shame.", "votes": 1 }, { "id": 974319, "postDate": "2020-08-17T22:10:19.857Z", "content": "@inversion you should also check for wheat detection and landmark retrieval 2020, they will also terminate soon
", "rawMarkdown": "@inversion you should also check for wheat detection and landmark retrieval 2020, they will also terminate soon", "votes": 2 }, { "id": 974400, "postDate": "2020-08-18T00:11:06.220Z", "content": "We are checking well beyond just this competition. :-)
", "rawMarkdown": "We are checking well beyond just this competition. :-)", "votes": 14 }, { "id": 976895, "postDate": "2020-08-19T06:53:36.640Z", "content": "I'll be really disappointed if the yikououbao isn't disqualified, look at the medals they have taken. They took in Melanoma and now are set for gold in Wheat. Seriously, the most brazen of groups.
", "rawMarkdown": "I'll be really disappointed if the yikououbao isn't disqualified, look at the medals they have taken. They took in Melanoma and now are set for gold in Wheat. Seriously, the most brazen of groups.", "votes": 2 }, { "id": 977422, "postDate": "2020-08-19T13:17:28.057Z", "content": "Thanks @inversion. The yikououbao account does look fishy. Always teaming up with noobs and only one repeated team member (i.e. yauwaiyim) etc. and there is a deleted account in the ALASKA2 competition among his/her team members.
", "rawMarkdown": "Thanks @inversion. The yikououbao account does look fishy. Always teaming up with noobs and only one repeated team member (i.e. yauwaiyim) etc. and there is a deleted account in the ALASKA2 competition among his/her team members.\n" } ] }, { "id": 973444, "postDate": "2020-08-17T09:58:00.747Z", "content": "I think they have a bunch of accounts for public leaderboard probing. What for? - It's still a mystery for me but I think that they will not score well on private part.
\nAny \"normal\" account will not have enough \"resources\" to conduct more than 5-10 experiments per day. But the number of accounts involved shows that they are doing 50-100 probes daily. With constant daily LB progress it could be only public probing.
\nThere is a \"question\" in my theory. Why they are not doing single record probing with low score and not using one or two accounts for \"public lb promotion sell\"?
", "rawMarkdown": "I think they have a bunch of accounts for public leaderboard probing. What for? - It's still a mystery for me but I think that they will not score well on private part.\n\nAny \"normal\" account will not have enough \"resources\" to conduct more than 5-10 experiments per day. But the number of accounts involved shows that they are doing 50-100 probes daily. With constant daily LB progress it could be only public probing.\n\n---\n\nThere is a \"question\" in my theory. Why they are not doing single record probing with low score and not using one or two accounts for \"public lb promotion sell\"? ", "votes": 3, "replies": [ { "id": 973459, "postDate": "2020-08-17T10:05:02.027Z", "content": "They had a few years worth of comps on their website. Years. Each comp had 25-100+ \"students\" registered in it. The operation is bigger than we know.
", "rawMarkdown": "They had a few years worth of comps on their website. **Years**. Each comp had 25-100+ \"students\" registered in it. The operation is bigger than we know.", "votes": 1 }, { "id": 973464, "postDate": "2020-08-17T10:09:41.263Z", "content": "Let's monitor the private LB. if a team get 0.98 or 0.99 in private without sharing their method I won't be convinced.
", "rawMarkdown": "Let's monitor the private LB. if a team get 0.98 or 0.99 in private without sharing their method I won't be convinced.", "votes": 1 }, { "id": 973484, "postDate": "2020-08-17T10:23:20.263Z", "content": "It gave them a short window to advertise to gullible people not knowing that there is a difference between public and private LB.
\n\"BUY A THIRD PLACE ON KAGGLE COMPETITION WITH OVER 3000 CONTESTANTS NOW FOR ONLY 500 DOLLARS\"!
", "rawMarkdown": "It gave them a short window to advertise to gullible people not knowing that there is a difference between public and private LB.\n\n\"BUY A THIRD PLACE ON KAGGLE COMPETITION WITH OVER 3000 CONTESTANTS NOW FOR ONLY 500 DOLLARS\"!", "votes": 3 } ] }, { "id": 973429, "postDate": "2020-08-17T09:46:54.733Z", "content": "One of them in the money zone in this competition. He may recover his payment. More surprisingly this is his 1st competition.
", "rawMarkdown": "[One of them](https://www.kaggle.com/chenggj) in the money zone in this competition. He may recover his payment. More surprisingly this is his **1st competition**. ", "votes": 3, "replies": [ { "id": 973442, "postDate": "2020-08-17T09:57:00.700Z", "content": "No need to worry about the top ones. I believe that they'll be deleted.
\nI worry more about the ones hide in silver/bronze whom are less obvious. They take their first medal in this competition and next time they are no more novices so people won't doubt them.
You don't have to be a prize winner for Kaggle to require you to submit your code for verification of your LB position. Kaggle - why not select sets of medal-winning accounts and require code submission?
", "rawMarkdown": "You don't have to be a prize winner for Kaggle to require you to submit your code for verification of your LB position. Kaggle - why not select sets of medal-winning accounts and require code submission?", "votes": 4 }, { "id": 970828, "postDate": "2020-08-14T21:04:01.863Z", "content": "This is sad to see…
", "rawMarkdown": "This is sad to see...", "votes": 4 }, { "id": 974463, "postDate": "2020-08-18T00:34:59.413Z", "content": "It's time to remove cheaters out. Let's keep Kaggle clean. Respect to all who did job, i hope you will get your medals instead of fake accounts and cheaters.
", "rawMarkdown": "It's time to remove cheaters out. Let's keep Kaggle clean. Respect to all who did job, i hope you will get your medals instead of fake accounts and cheaters. ", "votes": 1, "replies": [ { "id": 974477, "postDate": "2020-08-18T00:42:30.457Z", "content": "They are still on the LB, but Kaggle is obviously looking into it. I hope that when they delete them, people advance up to fill in the freed spots so that those who worked so hard can get the medals they deserve.
", "rawMarkdown": "They are still on the LB, but Kaggle is obviously looking into it. I hope that when they delete them, people advance up to fill in the freed spots so that those who worked so hard can get the medals they deserve.", "votes": 3 } ] }, { "id": 973965, "postDate": "2020-08-17T16:26:37.130Z", "content": "It seems like this would be an interesting premise for a future competition. For those who might be cheating, the irony would be delicious.
", "rawMarkdown": "It seems like this would be an interesting premise for a future competition. For those who might be cheating, the irony would be delicious.", "votes": 1 }, { "id": 970347, "postDate": "2020-08-14T11:29:15Z", "content": "Usually I see suspicious novice accounts teamed up. But this time … they choose to solo => take much more places
\nMaybe they're just served to aid their main account to get a top place. Like the deleted accounts.
This seems clearly like private sharing submissions
\n@philculliton , @juliaelliott you may need to investigate here.
\nIn case of private sharing, that would be unfair for honest people who fall behind.
How many of this kind of people were removed in top? We moved from 156 to 140, so 16 were removed above us and i am wondering how many teams of this kind are affected
", "rawMarkdown": "How many of this kind of people were removed in top? We moved from 156 to 140, so 16 were removed above us and i am wondering how many teams of this kind are affected ", "votes": 2, "replies": [ { "id": 976987, "postDate": "2020-08-19T08:04:33.560Z", "content": "I moved from 379 to 328 (From No-medal zone to medal zone), a total of 51 places.
\nEarlier there were approx 3410 participants and now it's 3319 which says a total of 91(approx.) has been removed so far.
\nIt feels like our hard work is getting rewarded and cheaters are getting punished.
\nThanks @inversion
Dear @group16,
\nI hope the medal farm came with a money-back guarantee; in the end the \"Magnificent 7\" didn't get their medals…
\n| # | \n△pub | \nTeam Name | \n
|---|---|---|
| 2336 | \n-2326 | \nDLWangYP | \n
| 1551 | \n-1548 | \ncheng | \n
| 2296 | \n-2289 | \nXue Yang 0827 | \n
| 2317 | \n-2311 | \n幼稚园接悟空 | \n
| 1386 | \n-1378 | \nbirdfly123 | \n
| 1047 | \n-1033 | \nWANGXUEQING618 | \n
| 2335 | \n-2322 | \nzengchan | \n
You can easily get to them through the public LB :)! They didn't have the greatest success let's say….
", "rawMarkdown": "You can easily get to them through the public LB :)! They didn't have the greatest success let's say....", "votes": 2 }, { "id": 975219, "postDate": "2020-08-18T07:46:57.753Z", "content": "The shake up was amazing wasn't it! This is my first kaggle competition and I have been stunned at the change from public LB to private LB.
\nI really have got a huge amount of respect now for people like @cdeotte who kept repeating \"trust the cv\" and don't overfit the public LB. Now I get it! :)
", "rawMarkdown": "The shake up was amazing wasn't it! This is my first kaggle competition and I have been stunned at the change from public LB to private LB.\n\nI really have got a huge amount of respect now for people like @cdeotte who kept repeating \"trust the cv\" and don't overfit the public LB. Now I get it! :)", "votes": 3 }, { "id": 976879, "postDate": "2020-08-19T06:41:37.610Z", "content": "mahahaha . Magnificant 7. Super nice wording! ;)👍
", "rawMarkdown": "mahahaha . Magnificant 7. Super nice wording! ;)👍", "votes": 1 }, { "id": 976893, "postDate": "2020-08-19T06:49:18.680Z", "content": "Although the Magnificant 7 got taken out, the Fearless 5 live on. They are re-loading their PayPal, CashApp, and other wallets, and with the help of more \"classes\" will be back for another comp.
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F69253%2F65ca872da15fb8306cc92cf7ba7fe8fb%2FScreen%20Shot%202020-08-19%20at%202.47.05%20AM.png?generation=1597819744758309&alt=media\")
Fearless Five.. another one ; ) and the only possiblity to still smile about it.
\nI guess in our opinion they have to be removed from the LB.
\nAny other nice findings, dear Signal?
\nAh btw: Please start to publish Notebooks with this well defined wording for groups ;)!
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2F0be78abe944f3136f61003fe19479ee8%2F_20200819103926.png?generation=1597826384637393&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F456596%2Ffa8e3e10d6838a77066faad28941dd2f%2F_20200819104332.png?generation=1597826644223976&alt=media\")
You just missed this one, I believe they're from the same group of \"Fearless Five\" and they are already experts 😄
", "rawMarkdown": "\n\n\n\nYou just missed this one, I believe they're from the same group of \"Fearless Five\" and they are already experts 😄", "votes": 1 } ] }, { "id": 973909, "postDate": "2020-08-17T15:39:18.897Z", "content": "@group16 This is an interesting find and I think that Kaggle should make sure this don't happen.
", "rawMarkdown": "@group16 This is an interesting find and I think that Kaggle should make sure this don't happen.", "votes": 2 }, { "id": 973381, "postDate": "2020-08-17T09:20:23.240Z", "content": "How many competitions have they (and also like them) entered ? And how many people who really worked hard for those competitions, missed their actual achievement and said , \"Oh, God. I don't worth it\"😳 ?
", "rawMarkdown": "How many competitions have they (and also like them) entered ? And how many people who really worked hard for those competitions, missed their actual achievement and said , \"Oh, God. I don't worth it\"😳 ? ", "votes": 2 }, { "id": 973186, "postDate": "2020-08-17T06:46:50.227Z", "content": "Is User data and what comps and teams (and teammates) they have been in, in a public API? If so we can just build a node graph, run some basic graph stats like (centrality, etc) and find out the entire \"network\" of these accounts.
", "rawMarkdown": "Is User data and what comps and teams (and teammates) they have been in, in a public API? If so we can just build a node graph, run some basic graph stats like (centrality, etc) and find out the entire \"network\" of these accounts.", "votes": 2 }, { "id": 973713, "postDate": "2020-08-17T13:25:30.110Z", "content": "from kaggle.garbage import garbage !!!
\ngarbage.collect()
So boring.If you spend the time and energy on your research, you can win the CVPR 2021 Best Paper Award.😴😴😴😴😴😴😴😴
", "rawMarkdown": "So boring.If you spend the time and energy on your research, you can win the CVPR 2021 Best Paper Award.😴😴😴😴😴😴😴😴", "votes": -13 }, { "id": 984708, "postDate": "2020-08-25T08:32:12.233Z", "content": "Did you try search answer in-home assignment services?
", "rawMarkdown": "Did you try search answer in-home assignment services?", "votes": -1 }, { "id": 974098, "postDate": "2020-08-17T18:12:26.287Z", "content": "Might be all from some circle or selling submission for some amount just to get into silver etc.
\nIt looks fishy but. Don't know but I hope kaggle can look up into this and make competitions more truthful.
Classic Kaggle, allowing this anonymous users to spam submissions until they get right submission 👌
", "rawMarkdown": "Classic Kaggle, allowing this anonymous users to spam submissions until they get right submission 👌" }, { "id": 973415, "postDate": "2020-08-17T09:35:01.453Z", "content": "wait and see how they score after 15 hours
", "rawMarkdown": "wait and see how they score after 15 hours" }, { "id": 973385, "postDate": "2020-08-17T09:21:47.537Z", "content": "This is super fishy, and also quite clear & obvious what's going on.
\nWe definitely should write a notebook about it - will get well-deserved-instant gold ;)
If so, I wonder how far that 'medal farm' process is automated. Is it still interaction of several humans players or more a kind of an algorithm that runs the same model, then applies minor changes to submissions to emulate diversity and finally submits them.
", "rawMarkdown": "If so, I wonder how far that 'medal farm' process is automated. Is it still interaction of several humans players or more a kind of an algorithm that runs the same model, then applies minor changes to submissions to emulate diversity and finally submits them." }, { "id": 970337, "postDate": "2020-08-14T11:19:07.043Z", "content": "multiple accounts 🤔🤔🤔🤔🤔🤔
", "rawMarkdown": "multiple accounts 🤔🤔🤔🤔🤔🤔" }, { "id": 970328, "postDate": "2020-08-14T11:13:27.840Z", "content": "Considerable Insight👍There seems to be something fishy going on.
", "rawMarkdown": "Considerable Insight👍There seems to be something fishy going on." }, { "id": 970324, "postDate": "2020-08-14T11:08:18.513Z", "content": "Could be someone at the top in the LB.
\nCreating multiple accounts probably to get around 'max submissions' or 'gpu hours'.
", "rawMarkdown": "Could be someone at the top in the LB.\n\nCreating multiple accounts probably to get around 'max submissions' or 'gpu hours'.", "replies": [ { "id": 970326, "postDate": "2020-08-14T11:10:43.867Z", "content": "That's a possibility. I honestly think these are medal farms.
\nNotice for example that dlwangyp and zengchan also finished at very similar positions in the humpback whale competition. These are the first accounts I've checked.
", "rawMarkdown": "That's a possibility. I honestly think these are medal farms.\n\nNotice for example that dlwangyp and zengchan also finished at very similar positions in the humpback whale competition. These are the first accounts I've checked.", "votes": 1, "replies": [ { "id": 972802, "postDate": "2020-08-16T20:39:40.583Z", "content": "Not a single post, comment etc from any of them. They just show up and jump to top of leaderboard.
", "rawMarkdown": "Not a single post, comment etc from any of them. They just show up and jump to top of leaderboard." } ] }, { "id": 970336, "postDate": "2020-08-14T11:18:26.953Z", "content": "Also 4 of those accounts joined kaggle 2 years ago and have been active this month. Coincidence ??
\nDoesn’t look like it.
", "rawMarkdown": "Also 4 of those accounts joined kaggle 2 years ago and have been active this month. Coincidence ??\n\nDoesn’t look like it." }, { "id": 970446, "postDate": "2020-08-14T13:11:16.937Z", "content": "Dear @group16,
\nCould you explain what a \"medal farm\" is? I am wondering why would somebody would want to gain a medal on what is, essentially, a throwaway account?
\nAll the best,
\ncarl
These throwaway accounts do become useful once they have one or two medals. I think those accounts are then being sold. So either it is 1 person trying to get medals on multiple accounts or some group where people have to pay for high-scoring submissions.
", "rawMarkdown": "Hi @carlmcbrideellis,\n\nThese throwaway accounts do become useful once they have one or two medals. I think those accounts are then being sold. So either it is 1 person trying to get medals on multiple accounts or some group where people have to pay for high-scoring submissions. ", "votes": 15 }, { "id": 970479, "postDate": "2020-08-14T13:52:59.237Z", "content": "Dear @group16,
\nWow, I have been overly naïve. It is evident that there are a good number of, let us say, 'enthusiastic' individuals on kaggle (in all of the sections, the most overt of which, sooner or later, usually end up becoming a 404). But the idea that people could be selling kaggle accounts pre-seeded with medals, or others buying quality submission.csv files, is something that was beyond my imagination. All in all this is indeed a sorry state of affairs.
All the best,
\ncarl
@group16 Finally someone understands what's happening, you're thinking in the right direction. I felt bad when seeing guys being naïve when facing at this kind of issue in other competitions, some people even tried to justify for them.
", "rawMarkdown": "@group16 Finally someone understands what's happening, you're thinking in the right direction. I felt bad when seeing guys being naïve when facing at this kind of issue in other competitions, some people even tried to justify for them.", "votes": 3 }, { "id": 970928, "postDate": "2020-08-15T02:35:16.127Z", "content": "oh dear, i thought acc selling only exists in games
", "rawMarkdown": "oh dear, i thought acc selling only exists in games", "votes": 1 }, { "id": 970968, "postDate": "2020-08-15T03:49:00.197Z", "rawMarkdown": "", "votes": 1, "isDeleted": true } ] }, { "id": 970501, "postDate": "2020-08-14T14:08:09.170Z", "content": "It's surprising! I have seen many people who buy a top-ranked online game account spending a lot of money. An account sale arising in kaggle is surprising! In terms of a positive view, it implies that kaggling is worth much! But, I suggest that the host be cautious about dealing with multiple accounts to foster the fair game by making competitors trust and accept their results accordingly at the end of this game.
\nlet's enjoy kaggle
Do you really think kaggle is worth anything if it is dominated by fraudulent activity? All this work, often money and energy spent to get a decent ranking only to be pushed down by medal farms? There are other data competitions out there.
\nKaggle should require an explanation of the methods used for the top-N positions and originality in general should be rewarded, no explanation means no top-ranking, no originality but just brute force means a penalty, novelty should be rewarded. Duplicate accounts means an immediate ban, private sharing means an immediate ban, no data blending, etc.
\nNow it is primarily a shit show of people copying each other's notebooks and even submission files, and only a few very generous people sharing their work (people who should in my view win this competition based on merit alone). I see people happily show casing how to \"blend\" submissions (which is COMPLETELY useless for the competition sponsor). Why not ONLY share notebooks AFTER the competitions? This forces people to innovate and with the requirement of explainability and originality you weed out most of the bad apples.
\nIn the mean time we have collectively used enough energy running TPU's and GPU's to power a decently sized city for a week or so.
\nI learned a lot regarding pytorch and tensorflow in the last week, and I learned how amazingly fast TPU's are so I am grateful for that but this gives very bad aftertaste.
\n[edit]
\nThis is downvoted? Incredible. It is obvious that Kaggle is flawed in it's current form. The amount of redundancy in approaches is incredible, the amount of energy wasted mind boggling and being able to publish submission files and even winning approaches before the competition ends is an invitation to unfairness and a guarantee that the diversity in final approaches is LIMITED. The majority of people basically MIMICKED eachother, what is the point? To learn? I am pretty sure we can do that on data sets that are a fraction of the size. People have used hundreds of TPU hours repeating eachothers work, in the mean time the competition sponsor has learned nothing new except that running incredibly expensive TPU's on huge models pays off..woohoo.
Or they made you think that way ;)
", "rawMarkdown": "Or they made you think that way ;)", "votes": 2 }, { "id": 972549, "postDate": "2020-08-16T16:19:16.890Z", "content": "All are offenders, not only the one sharing (might be also same person)
", "rawMarkdown": "All are offenders, not only the one sharing (might be also same person)", "votes": 1 }, { "id": 972571, "postDate": "2020-08-16T16:39:45.833Z", "content": "@gunesevitan but why they want to create novice accounts to push his main account out of gold ? Why they do not use the highest subs themselves ?
", "rawMarkdown": "@gunesevitan but why they want to create novice accounts to push his main account out of gold ? Why they do not use the highest subs themselves ?" } ] }, { "id": 971626, "postDate": "2020-08-15T18:27:53.263Z", "rawMarkdown": "", "isDeleted": true }, { "id": 970940, "postDate": "2020-08-15T02:54:28.137Z", "rawMarkdown": "", "votes": -1, "isDeleted": true }, { "id": 970935, "postDate": "2020-08-15T02:48:00.180Z", "rawMarkdown": "", "isDeleted": true }, { "id": 972887, "postDate": "2020-08-17T00:08:22.770Z", "content": "Great work, thanks for the contribution
", "rawMarkdown": "Great work, thanks for the contribution" } ], "comments": [ { "id": 971492, "author_name": "Chris Deotte", "author_url": "", "post_date": "2020-08-15T15:21:39.683000", "content": "What a strange coincidence. All 7 of these teams jumped over me today. How can 7 teams working independently all make the same discovery (to increase their score) at the same time?
\n(Note that this happened in a span of a few hours and no high scoring public notebooks were published during this time).
\n
In the span of (maximum) 2 hours even :). What a coincidence indeed!
", "votes": 3, "replies": [] }, { "id": 971496, "author_name": "gao-hongnan", "author_url": "", "post_date": "2020-08-15T15:27:36.737000", "content": "@cdeotte I'm pretty new to these type of antics, what are they trying to achieve?
", "votes": 5, "replies": [] }, { "id": 971501, "author_name": "Shimei", "author_url": "", "post_date": "2020-08-15T15:37:08.360000", "content": "Obviously group behavior of the same organization
", "votes": 1, "replies": [] }, { "id": 971512, "author_name": "Serigne ", "author_url": "", "post_date": "2020-08-15T15:59:37.670000", "content": "Coïncidence ? Je ne crois pas… ^^
", "votes": 8, "replies": [] }, { "id": 971523, "author_name": "Ertuğrul Demir", "author_url": "", "post_date": "2020-08-15T16:17:35.123000", "content": "Expecting this is not gonna get noticed in the heart of the data science community, oh the irony…
", "votes": 14, "replies": [] }, { "id": 971661, "author_name": "Stanley Zheng", "author_url": "", "post_date": "2020-08-15T19:28:16.657000", "content": "@reighns They could be trying to sell accounts? I don't see any other way to benefit off of having multiple accounts with nearly identical placements…
", "votes": 1, "replies": [] }, { "id": 971711, "author_name": "Signal", "author_url": "", "post_date": "2020-08-15T20:23:14.233000", "content": "the Wang Cheng Yang cabal strikes again! People have put their soul into this comp, so unfair if there is any cheating or collusion going on. Private sharing is unfair compared to a legitimate team. Discouraging. Imagine being that person that just misses a gold, silver or bronze by one position.
", "votes": 8, "replies": [] }, { "id": 971804, "author_name": "YaGana Sheriff-Hussaini", "author_url": "", "post_date": "2020-08-15T22:59:56.560000", "content": "@cdeotte, since I discovered something called \"click fraud \" when I participated in the TalkingData AdTracking Fraud Detection Challenge, my distrust level for this kind of thing went up exponentially.
\nI agree with @serigne. Grosse coïncidence en effet. J'en doute.
", "votes": 3, "replies": [] }, { "id": 972089, "author_name": "from coffee import *", "author_url": "", "post_date": "2020-08-16T08:21:10.730000", "content": "@maggiemd, @philculliton, wouldn't this be worth a look at? There are several cases of Novice and not-Novice jumps probably using the same non-public submissions. But this specific team really is … an outlier considering the attribute \"audacity\"
\nThe irony of fraud detection not working in the world's best Data-Science community is almost worth an own notebook 😃
I am so glad you posted this Chris… I mean some people, including myself, work really hard alone (or in a team of 2) to get a medal and here comes this mafia stealing from us… I'm always 10 spots behind a medal, I really wonder why….. Thank you for raising this questions, maybe Kaggle staff wakes up….
", "votes": 1, "replies": [] } ] }, { "id": 970515, "author_name": "Rob Mulla", "author_url": "", "post_date": "2020-08-14T14:21:47.183000", "content": "Some similarities in when subs were made too.
\nThis shows subs from just the last 5 days. Only shows subs that increased LB score. Makes sense that a group would sub at the same time of day if they are coordinating… or could just be a coincidence.
\n
Thanks @robikscube, that's indeed a great indicator. It's insane how obvious they have made it…
\nThree of them are also collaborators on the same dataset: https://www.kaggle.com/wangxueqing618/tfrecord384x384
", "votes": 11, "replies": [] }, { "id": 970775, "author_name": "Nischay Dhankhar", "author_url": "", "post_date": "2020-08-14T19:21:04.980000", "content": "Wow.It explains everything.
", "votes": 2, "replies": [] }, { "id": 970925, "author_name": "Signal", "author_url": "", "post_date": "2020-08-15T02:27:08.827000", "content": "@robikscube wow, I'm convinced. I am sure Kaggle does basic things like check for accesses from the same IP address. Most people are smart enough to use VPN though. Kaggle should have its own Kaggle competition to detect fraud, and then use what it learns to improve.
\nOne basic way would be to charge………..I realize that has all kinds of problems, but it does stop people from creating basic farm accounts en masse. Another way would be to do basic things like cell # verification. I realize people can have multiple #'s or google voice, etc. But I mean, if you put enough controls in place it does help things.
", "votes": 5, "replies": [] }, { "id": 977003, "author_name": "Janek Idziak", "author_url": "", "post_date": "2020-08-19T08:20:02.537000", "content": "Out of curiosity, how do you get the data to create this plot?
", "votes": 0, "replies": [] }, { "id": 977382, "author_name": "Rob Mulla", "author_url": "", "post_date": "2020-08-19T12:53:51.257000", "content": "@janidziak, you look at the public leaderboard tab, there is a \"download raw data\" button you can click to get a CSV file of the submissions. It shows all submissions that caused teams to increase their public leaderboard score.
", "votes": 1, "replies": [] }, { "id": 979115, "author_name": "Janek Idziak", "author_url": "", "post_date": "2020-08-20T16:16:38.357000", "content": "Thank you @robikscube
", "votes": 0, "replies": [] } ] }, { "id": 972076, "author_name": "ZachhhBee", "author_url": "", "post_date": "2020-08-16T07:59:08.897000", "content": "I wish those are just more accounts for GPU/TPU time. But kaggle's reputation has been downfalling fast in my country because there are so called medal factories.
\nA medal factory sold medals by sharing private notebooks. Thousands of dollars for each one getting a medal with their private notebook in the cost of 20yuan/3dollars for VPN each month. Gold medals worth more. And people buying those medals will use kaggle's reputation and medal quality in trade for jobs/applying graduates in school etc. They damage kaggle's reputation, got their own benefit and then leave, possibly never come back.
\nMy teammate and I have devoted hours and days in this competition to improve our cv and climp the LB little by little. All of a sudden those novices just come up and surpass you. When the legend of medal factories happened, things can be very FRUSTRATED and MIND-FFFFING. SAD indeed. Hope kaggle will not be another \"game destroyed by cheaters\". I love this community.
\nI wish legend is just legend but not truth.
\n@philculliton , @juliaelliott, @Phil Culliton
", "votes": 28, "replies": [] }, { "id": 972601, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-16T17:12:00.580000", "content": "And this team is just LEGENDARY. (all screenshots are from public LB)
\nLiverpool:
\n
Alaska2:
\n
Panda:
\n
Wheat:
\n
Melanoma:
\n
Landmark2020:
\n
Yet Kaggle has done nothing to them. Is there any explaination?
", "votes": 19, "replies": [ { "id": 972609, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-16T17:18:16.753000", "content": "That makes some team above them suspicious 🙂
", "votes": 0, "replies": [ { "id": 973230, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T07:26:18.983000", "content": "@jielu0728 same profile as all the accounts I found and others posted.
\nNo activity, no posts, no comments, no bio, no followers, not following anyone. They just joined Kaggle and started climbing competitions. So blatant, in your face……..I mean no shame at all.
\nTop 1% in Wheat and Melanoma…………..about to take down medals if Kaggle does not do something about it.
", "votes": 2, "replies": [] }, { "id": 974950, "author_name": "", "author_url": "", "post_date": "2020-08-18T05:09:40.153000", "content": "Some of them does have skills. Pity to see the way using it…
", "votes": 0, "replies": [] } ] }, { "id": 973246, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2020-08-17T07:37:07.483000", "content": "Nothing suspicious at all…
\n
@group16 lol 4 submissions. All of those team mates are in on it too, these aren't random team meetups. It seems to have been going on for at least 2 years, but probably more based on what I saw on the website before @lifeifei38 took it down.
", "votes": 3, "replies": [] }, { "id": 973258, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2020-08-17T07:47:22.577000", "content": "Exactly. The low submission count & the team of novices is what made them stood out in the first place. Both of them have a abnormal low amount of submissions to score that high, which indicates that other teams were involved as well.
\nIt is weird that this team is so successful, and that 4-5 of the team members haven't even bothered to get the contributor tier (which would require them to verify themselves).
", "votes": 2, "replies": [] }, { "id": 973287, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T08:03:24.513000", "content": "@group16 so they must not be aspiring Data Scientists. They must just be mules, farm accounts, and by not filling in any bio or info, that probably makes the account more valuable, as its shiny and new. How is it, a few of us pull this info together in minutes and nothing gets done about these people? They are seriously going to take medals, possibly gold, in both Wheat and Melanoma in the next 15 hours. I'm really raging about this, i seriously wish we had access to data to build a node graph with edges, just throw in the dozen or two names we have already doxed and see all who is fruit of the poisonous tree.
", "votes": 1, "replies": [] }, { "id": 973290, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2020-08-17T08:06:54.453000", "content": "It should be possible to do graph analysis based on the Kaggle Meta-dataset. It's going to be quite a bit of work though, but it would indeed be interested to see. Pretty sure we can create a large graph that connects all of these accounts together.
", "votes": 0, "replies": [] }, { "id": 973386, "author_name": "Bram Steenwinckel", "author_url": "", "post_date": "2020-08-17T09:22:13.010000", "content": "Sounds like a plan
", "votes": 0, "replies": [] } ] } ] }, { "id": 970355, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2020-08-14T11:34:33.130000", "content": "Kaggle should make sure to have some detection systems in place. I found something very suspicious during the Liverpool competition as well. I never checked up on it, but appears one of them have gotten silver medals (the other probably used multiple accounts and got banned for that). Both of them are now in a team together in the wheat detection competition…
\n
There isn't any doubt they cheated. Both accounts should be banned from Kaggle.
", "votes": 4, "replies": [] }, { "id": 970412, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-14T12:46:18.643000", "content": "
Enjoy!
", "votes": 16, "replies": [] }, { "id": 973178, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2020-08-17T06:43:43.520000", "content": "This is so funny and sad 😄
", "votes": 0, "replies": [] } ] }, { "id": 972892, "author_name": "Waylon Wu", "author_url": "", "post_date": "2020-08-17T00:20:28.593000", "content": "I guess this can be related to https://edu.angtk.com/course/35 ? This competition worths $1000?
\n
Edit: the website is down now.
", "votes": 14, "replies": [ { "id": 972906, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T00:59:43.440000", "content": "That website mentions some people who \"teach\" by name:
\nZhang Peng
\nGuo Junlin
\nChen Zidong
\nLi Xiaoyu
I assume these people are on Kaggle?
", "votes": 9, "replies": [] }, { "id": 972926, "author_name": "", "author_url": "", "post_date": "2020-08-17T01:27:58.580000", "content": "", "votes": 0, "replies": [] }, { "id": 972943, "author_name": "", "author_url": "", "post_date": "2020-08-17T01:43:44.633000", "content": "", "votes": 4, "replies": [] }, { "id": 972958, "author_name": "", "author_url": "", "post_date": "2020-08-17T02:02:29.273000", "content": "", "votes": 3, "replies": [] }, { "id": 972965, "author_name": "", "author_url": "", "post_date": "2020-08-17T02:15:46.937000", "content": "We are just a teaching website. All teaching materials can be provided to the organizers for viewing. All of them are open baseline procedures. And we don't offer to know what to share in private. This video, we can put on the youtube video website, just explained some basic knowledge of public baseline!
", "votes": -19, "replies": [] }, { "id": 972969, "author_name": "", "author_url": "", "post_date": "2020-08-17T02:20:32.367000", "content": "I can be sure that our students have participated in this competition. They all use the open baseline to learn. But not in the gold medal area. All methods are based on the discussion group here or in notebooks.
", "votes": -17, "replies": [ { "id": 973012, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T03:28:25.350000", "content": "And they pay $1000? to people like you who have never even posted on Kaggle until today? Never entered a comp? Do all the instructors have accounts with 0 competitions entered and no posts in the Kaggle community? I mean, you realize how this looks right? You just identified yourself as one of the instructors. Yet you also have an account @fffeiyee which has entered comps with other \"instructors\" on that site such as @dongxiaoming . You realize having multiple accounts is against the rules right?
", "votes": 16, "replies": [] }, { "id": 973026, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T03:51:21.050000", "content": "Under \"Student News\" for the Kaggle Melanoma class (offered at $1000), are two students……….and there names just happen to be……….wait for it………. Wang and Yang:
\n
Thank you for your question. Let me answer it
\nFirst of all, I will clarify our knowledge website teaching and Q & A. The teacher is part-time! The screenshot is the video content of this website, and the video I explained is the video for basic teaching. We have no right to disclose the teacher's account number. They just record some basic teaching videos on a part-time basis. We have not organized students to participate in the competition. They will ask questions if they have questions, but they will not provide us with their accounts.
\nAnd the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this.
", "votes": -7, "replies": [] }, { "id": 973033, "author_name": "", "author_url": "", "post_date": "2020-08-17T04:00:59.853000", "content": "Watching the same basic video doesn't mean sharing it in private. This video is not related to match scoring. Just use the baseline to explain the basis of deep learning. I said, I will release these videos tomorrow and then youtube, we are registering and uploading!
", "votes": -11, "replies": [] }, { "id": 973054, "author_name": "Gunes Evitan", "author_url": "", "post_date": "2020-08-17T04:41:41.810000", "content": "\n\nAnd the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this.
\n
How much does it cost? Even though it doesn't cost that much, it looks fishy.
", "votes": 2, "replies": [] }, { "id": 973203, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T07:02:50.917000", "content": "@lifeifei38 Why did you just take the entire web site down? You must be the/one of the kingpins. Making money off selling Kaggle comps and you don't even participate. No doubt lifefei38 is a fake account, it has no activity, accept this thread………I hope your real identity is discovered.
", "votes": 3, "replies": [] }, { "id": 973236, "author_name": "", "author_url": "", "post_date": "2020-08-17T07:28:32.993000", "content": "We are really sorry that we used an incorrect word in the sentence \"Our actual cost is lower than this.\" We wanted to say \"The actual price of our lesson is lower than the price showing in our website.\" This is a language problem, we carelessly used the word \"cost\" by mistake. The lessons and their prices on our website is just for show, while when a student want to attend our lesson, he/she contact us by instant messaging application and we will further discuss the price concerning his/her needs and open the suitable lesson(s) for him/her.
\nWe will upload lessons to Youtube tomorrow. The lessons are in Chinese but we will use automatic or human translation to add English subtitles and let you (also Kaggle and other Kagglers) know that our lessons
\ninclude only basic explanations, but not the hints for improving the score of the competition.
\nA lot of cross-disciplinary beginners for Kaggle have difficulties taking part in the competitions because they are unfamiliar with the operation of Kaggle or have questions concerning the basic knowledge, this is where our lessons helps.
I just explain the video content of our website.There are a lot of similar videos. On youtube, you can go there and grab account. We have an obligation to protect our web account information
", "votes": -11, "replies": [] } ] }, { "id": 972975, "author_name": "", "author_url": "", "post_date": "2020-08-17T02:27:34.333000", "content": "We can put all three videos of this course on YouTube tomorrow, August 18. And this video hasn't been updated half a month ago. We only provide the most basic baseline learning lectures, not the sharing of competition skills.
", "votes": -2, "replies": [] }, { "id": 973008, "author_name": "ZachhhBee", "author_url": "", "post_date": "2020-08-17T03:14:08.353000", "content": "Yeah, you don't offer to know. We know something called social media with digital pay in it.
\nThanks for your explanation to clear our doubts. Your four videos each len 40 minutes worth 1000dollars. There must be sth really serious and hard stuff for us to understand. So for more dicretion and easy understanding, why don't u guys just share your teachers' and students' real kaggle profile page and email for the organizer to investigate. We see that you have many teachers and 36 students, right? They must be on kaggle and submitted baselines, right? So they can be evaluated? And your teachers must be very experienced and good on kaggle so people pay 1000 dollars to listen for their less than 3 hours speech, right? GrandMaster maybe?
\nI'm very happy to see those genius novices on kaggle who will in the future do great job for AI development and benefit more people with their amazing skills. Those people are hopes for AI's next stage fast development.
\nThanks again for your work in teaching and your help to thrive the kaggle community by teaching others.
", "votes": 14, "replies": [] }, { "id": 973020, "author_name": "", "author_url": "", "post_date": "2020-08-17T03:43:59.523000", "content": "First of all, thank you very much for your reasonably requesting us to clarify the whole matter.
\nOur students just come to study, I don't know their account number. In addition, we have online basic Q & A, as well as basic questions of deep learning and machine learning on this website. And the prices marked on the website are just for display. You can try to pay but the payment will fail. Our actual cost is lower than this. Many teachers are only responsible for explaining the basic knowledge concerning the baseline and answering questions. In the future, we will organize teachers to share some notebooks, but now we have no plan to disclose personal information as we need to ask for personal consent. In addition, we can not control whether the students will discuss in other places after watching the course. We are just an online teaching and answering website, not an actual university with regulations for students.
", "votes": -4, "replies": [] }, { "id": 973063, "author_name": "", "author_url": "", "post_date": "2020-08-17T04:55:19.057000", "content": "", "votes": 0, "replies": [] }, { "id": 973066, "author_name": "", "author_url": "", "post_date": "2020-08-17T04:59:13.750000", "content": "", "votes": 0, "replies": [] }, { "id": 973096, "author_name": "ZachhhBee", "author_url": "", "post_date": "2020-08-17T05:27:41.373000", "content": "You know you just said that you guys really take money from this. Where did those happen, where did your guys get paid since it't not on the official webiste? How many of those secret payments relating to private \"baseline explanaion\" & \"Q&A\" etc happened? We don't know; The kaggle officials don't know; You may or may not know.
\nSeriously, is this account your offical account on kaggle? Should people focus this account if he/she wants to learn in your \"community\"?
\nDude, this thread is gonna be very famous and controversial. A really famous one thanks to you from an account only in charge of discussion under this thread with thousands of words revealing what you have done. And yet you dodge key questions- real kaggle profiles of your precious teachers must also be private because they are that kind of online teachers serving a certain group of \"student\" who paid well and don't like others to see them? Right, a bit strange for teachers on a ML&DL teaching webisite maybe? Cause teachers elsewhere online just want more people to know them.
\nI think I'm not gonna to put any more time into this shit. Keep talking and let more people to know this thread, to know what your guys are and what you've done and your clarification. Good job and keep working! Maybe I'll see this thread in next BigDataDigest news - finally someone behind those genius novices come up, a new novice account with zero competition experience but not GrandMaster could be disappointed. But still, Great source! Great ad! Good for your business!
", "votes": 8, "replies": [ { "id": 973138, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T06:07:17.540000", "content": "Want to be sick to your stomach? I just went to their M5 \"class\" page, and it shows \"New Students\". It shows this for each class. Each class is full of brand new kaggle accounts, with no activity, no post history, just enter the competition, with other accounts who also are new and have no history. and get high scores. Here is just what I manually looked up on M5. Also realize each of these is teamed with other people, who also are newbies, and those people team with people, and it goes on and on…….
\nScraped from Link for M5 competition “class” “Newest Students” at this link: https://translate.google.com/translate?hl=en&sl=zh-CN&u=https://edu.angtk.com/course/32/notes%3Fsort%3Dlike%26task%3D465%26selectedCourse%3D0&prev=search&pto=aue
\nStudent 1
\naustin4052 aka Austin Kang, Appears in Kaggle for the first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 559 out of 5500+ people just barely missing a bronze.
\nTeamed with
\n 红警free ( liyishuolys)
\n Zhengbo ZHOU (curryzhou)
\n Ningcan Sun (ningcansun)
\n Jianzhong Tian (jianzhongtian)
Student 2
\nEric_Tonser Appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 551 out of 5500+ people medaling a bronze.
\nTeamed with
\n 不正常人类研究中心 (sjbowen)
\n 小鱼仔 (xiaoyuzai)
\n Cherish_L(cherishl)
\n KeepingFly (keepingfly)
Student 3
\nJesse (Kaggle info unknown)
Student 4
\nSun Ningcan appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 559 out of 5500+ people just barely missing a bronze.
\nTeamed with
\n 红警free ( liyishuolys)
\n Zhengbo ZHOU (curryzhou)
\n Austin Kang
\n Jianzhong Tian (jianzhongtian)
Student 5
\nHxysophy (Kaggle info unknown)
Student 6
\nKoko (Kaggle info unknown)
Student 7
\nMadokazu (Kaggle info unknown)
Student 8
\nKikkk which looks awfully similar to kikikkk which appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 1325 out of 5500+
\nTeamed with:
\n Henry Tang (henryshtang)
\n Mengton Xu (mengtongxu)
\n Xiang Anqi (xianganqi)
Student 9
\nSinner0612 (Kaggle info unknown)
Student 10
\nericyuu (Kaggle info unknown)
Student 11
\nSnowmeli (snowmeli) appears in Kaggle for first time, never posts a single thing, still hasn’t. Joins M5 Accuracy and gets 490 out of 5500+
\nTeamed with
\n машинное обучение (a17186819035)
\n Hubert He (anuhw0228)
\n WeiHao Li (josephliweihao)
Student 12
\nanuhw (Kaggle info unknown)
Student 13
\nliweihao (liweihao) interesting that it is the same “name” as the person I just mentioned above WeiHao Li (josephliweihao)
Student 14
\nTaohuahuahuanyugan (Kaggle info unknown)
Student 15
\nWoww (Kaggle info unknown)
Student 16
\nJRiY (Kaggle info unknown)
Student 17
\nPowdererer (Kaggle info unknown)
Student 18
\nquanbingtao (Kaggle info unknown)
Student 19
\nchiaraduan (Kaggle info unknown)
Student 20
\nMaster Yi Star
Google's history page clearly indicates that what we do is EDA and simple baseline submission. It doesn't involve any scoring skills and tricks. Is it against the rules here to submit several notebooks and not get medals? In addition, it is against the law to disclose information that we don't want to control others?
That's definitely against the rules. It's called private sharing. I remember things being shared on a slack channel during the instant gratification competition, which was also classified as private sharing.
", "votes": 16, "replies": [] }, { "id": 973167, "author_name": "Gunes Evitan", "author_url": "", "post_date": "2020-08-17T06:34:51.037000", "content": "If you did nothing against the rules, you can share the kaggle profiles of your teachers and students.
", "votes": 6, "replies": [] }, { "id": 973172, "author_name": "", "author_url": "", "post_date": "2020-08-17T06:38:40.203000", "content": "The students just came here to study, and they didn't tell us what the account number is. We only answer questions, such as their registration or notebook questions. And we can also see that we use the open source baseline of kaggle to do some more detailed technical explanation.
", "votes": -5, "replies": [] }, { "id": 973177, "author_name": "", "author_url": "", "post_date": "2020-08-17T06:43:23.997000", "content": "", "votes": 0, "replies": [] }, { "id": 973179, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T06:43:57.340000", "content": "Amazing! Here are some of your students in Melanoma.
\nWhats fascinating is that all your students join Kaggle. Never post, never comment, have no activity. They join large teams. Everyone in the team is the same, they are all new, no activity. The teams all score near the top, some medal.
\n\nStudent 1
\nGawain Gan (gawaingan) Top 11% M5 Forecasting, Top 12% in Melanoma, Top 26% in Prostate
\nTeamed with (Melanoma)
\n Watmics (watmics)
\n Jiaqing Xie (godxie)
\n Sifan Wang96 (sifanwang96)
\n gw.w9 (guoweiw9)
Student 2
\nBochen (Kaggle info unknown)
Student 3
\nmath
Student 4
\nZhao Yujie (zhaoyujie123) Currently 209 in Melanoma. Like everyone in this team has done the same thing, new account, no posts, no comments, etc, etc.
\nTeamed with
\n Patrick (tangyaxu)
\n Du Shaohui (dushaohui)
\n Xinjian (xinjianli)
\n Noone (neyiii)
Student 5
\nJoe1022 (Kaggle info unknown)
Student 6
\nTangyaxu Patrick (tangyaxu). Mentioned in above team
Student 7
\nGw_wang (Kaggle info unknown)
Student 8
\nMadokazu (Kaggle info unknown) Mentioned in M5 list too
Student 9
\nSnowmeli (snowmeli) was in M5 and prostrate
Student 10
\nAnuhw (Kaggle info unknown)
Student 11
\nLiweihao (liweihao)
Student 12
\nChangbc777 Siyu Guan (changbc777) no comps entered yet, no activity
Student 13
\nJIAQINGXIE (godxie) mentioned above in team, and set to medal
Student 14
\nHobart01 (Kaggle info unknown)
Student 15
\nzzzjh (Kaggle info unknown)
Student 16
\nLiweihao (liweihao) close to someone in my other list , same “name” as the person I just mentioned above WeiHao Li (josephliweihao)
Student 17
\nXmetric (Kaggle info unknown)
Student 18
\nsmrqqqq (Kaggle info unknown)
Student 19
\nJasonHuang (Kaggle info unknown)
Student 20
\njiacheng_shi (Kaggle info unknown)
Student 21
\naskpatrickgor (Kaggle info unknown)
Sharing notebooks only with people that pay for it and making them not publicly available and known to everyone is the definition of private sharing.
", "votes": 6, "replies": [] }, { "id": 973207, "author_name": "", "author_url": "", "post_date": "2020-08-17T07:07:21.840000", "content": "What we use to teach is an public baseline in notebook.
", "votes": -1, "replies": [] }, { "id": 973326, "author_name": "Aziz_Belaweid", "author_url": "", "post_date": "2020-08-17T08:45:11.717000", "content": "You keep saying that you are teaching them using public kernels , what's your addon ? why would someone pay for public information they can go on kaggle and see public notebooks and if they have any questions I'm sure the author will answer them, my opinion is they're paying to get push on lb whether it's a submission file or leaks that you are probably not allowed to share, if not why would they pay you ? ….
", "votes": 9, "replies": [] } ] }, { "id": 973252, "author_name": "", "author_url": "", "post_date": "2020-08-17T07:40:09.493000", "content": "", "votes": 3, "replies": [] }, { "id": 973275, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T07:56:22.480000", "content": "I took many screenshots before @lifeifei38 took it down. I was using translate but many of the writings are pics not text.
\nHere is a dropbox with a few of the pics:
\nhttps://www.dropbox.com/sh/mev70eeh7tvln0g/AAAyGMupYhMTZ3lo1FIeZc7Ha?dl=0
It is very frustrating to see my position going down on LB so fast…
\nI am a beginner and this really is disappointing to see.
\nI am not against public kernels as long as they just give ideas ( some great public kernels have helped me throughout this competition…increasing my knowledge ), but when they directly give the submission file…it is very unfair..
Thank you everyone to be active on discussions and publishing kernels with new ideas.
", "votes": 11, "replies": [ { "id": 979138, "author_name": "Prateek Mishra", "author_url": "", "post_date": "2020-08-20T16:42:24.077000", "content": "why you are not showing up on LB ? I have seen you when competition was going on.
\nwhat was your rank ?
Is it possible to remove yourself from a comp once you have joined? I don't think so.
", "votes": 0, "replies": [] } ] }, { "id": 970572, "author_name": "Sxwat", "author_url": "", "post_date": "2020-08-14T15:19:01.517000", "content": "We need Kaggle competition \"Find fake account in Kaggle\" and it is important to find real main account to punish cheater (remove medals or bans or something else). In every online game we have cheaters, so in Kaggle, and it is always annoying
", "votes": 12, "replies": [ { "id": 970582, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-14T15:24:08.087000", "content": "I think Kaggle's algorithm to remove cheaters really needs to be improved
", "votes": 13, "replies": [] }, { "id": 974258, "author_name": "Bram Van Es", "author_url": "", "post_date": "2020-08-17T20:44:51.420000", "content": "Having said that, where is Kaggle in this thread?? …Never mind, they have arrived.
", "votes": 0, "replies": [ { "id": 974264, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T20:53:35.140000", "content": "The silence isn't comforting, but at the same time, this is a rather large can of worms that just got opened so I am sure it's quite a bit for them to deal with.
", "votes": 0, "replies": [] } ] } ] }, { "id": 970404, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-14T12:32:37.303000", "content": "From liverpool to Trends to Alaska to Wheat detection to this one to even landmark 2020. Kaggle has done little effort, even now some of those competitions have been finalized, those LB still contain many cheaters.
", "votes": 12, "replies": [ { "id": 973297, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T08:17:15.330000", "content": "yikououbao has taken 3 medals so far, about to take 2 more…..it's not just one team though. Some are more obvious than others, and you can certainly cross-reference the team members with other teams they have been in to uncover more of the conspirators, but it comes down to that there just aren't the right controls in place to prevent this sort of thing.
", "votes": 2, "replies": [] } ] }, { "id": 973654, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T12:38:03.653000", "content": "One thing I learned is that loading all of Users.csv and TeamMembership.csv from Kaggles Meta Dataset makes Gephi drop to its knees, lol.
If someone has some time an interesting Node Graph would be:
\nUsers.csv)TeamMemberships.csv)TeamMemberships.csv)Additionally, you would want to join the User id's you find in Step 2, using Users.csv as Labels. Same with Teams instead of just Team ID's you could join the ID's in TeamMemberships.csv with Team labels found in Teams.csv.
I think you will find that its a closed loop. The networks will only grow so far. All will contain the same types of users. Users with no bio, no comments, no posts, immediately entering competitions and getting high scores.
\nGood tools for doing this are Gephi (slow), NetworkX, Neo4j
\nGreat tools for doing this would be cuGraph (RAPIDS). I will do some myself, but I have not slept in over 24 hours, and I know time is not our friend when medals stand to be lost in a matter of hours.
A list of the Users mentioned in this thread is below:
\n\"Id\",\"UserName\",\"DisplayName\",\"RegisterDate\",\"PerformanceTier\"\n\"2607029\",\"dlwangyp\",\"DLWangYP\",\"12/12/2018\",\"0\"\n\"1834362\",\"chenggj\",\"cheng\",\"04/18/2018\",\"0\"\n\"5214874\",\"xueyang0827\",\"Xue Yang 0827\",\"06/02/2020\",\"0\"\n\"5581487\",\"yosefhuang\",\"yosef huang\",\"08/06/2020\",\"0\"\n\"2301868\",\"lg920810\",\"幼稚园接悟空\",\"09/29/2018\",\"0\"\n\"5259643\",\"birdfly123\",\"birdfly123\",\"06/08/2020\",\"0\"\n\"5582725\",\"zhaogongzuo2\",\"zhaogongzuo2\",\"08/06/2020\",\"0\"\n\"5413611\",\"marwafatto\",\"Marwa_Fatto\",\"07/04/2020\",\"0\"\n\"2592536\",\"zengchan\",\"zengchan\",\"12/08/2018\",\"0\"\n\"3859401\",\"silewang\",\"480073534\",\"10/15/2019\",\"0\"\n\"4198084\",\"yikououbao\",\"yikououbao\",\"12/13/2019\",\"2\"\n\"4360898\",\"eliowang\",\"Edward\",\"01/21/2020\",\"0\"\n\"5285253\",\"yauwaiyim\",\"Yau Wai Yim\",\"06/12/2020\",\"0\"\n\"5366857\",\"siqicao\",\"siqi cao\",\"06/25/2020\",\"0\"\n\"2000407\",\"austin4052\",\"Austin Kang\",\"06/17/2018\",\"0\"\n\"3101840\",\"liyishuolys\",\"红警free\",\"04/17/2019\",\"0\"\n\"3858729\",\"curryzhou\",\"Zhengbo ZHOU\",\"10/15/2019\",\"0\"\n\"5292127\",\"ningcansun\",\"Ningcan Sun\",\"06/13/2020\",\"0\"\n\"5346060\",\"jianzhongtian\",\"Jianzhong Tian\",\"06/22/2020\",\"0\"\n\"5321313\",\"tonserboyt\",\"TonserBoyT\",\"06/17/2020\",\"0\"\n\"2438897\",\"sjbowen\",\"不正常人类研究中心\",\"11/02/2018\",\"0\"\n\"4100810\",\"xiaoyuzai\",\"小鱼仔\",\"11/25/2019\",\"0\"\n\"4321861\",\"cherishl\",\"Cherish_L\",\"01/13/2020\",\"1\"\n\"5175992\",\"keepingfly\",\"KeepingFly\",\"05/27/2020\",\"0\"\n\"4672128\",\"kikikkk\",\"kikikkk\",\"03/15/2020\",\"0\"\n\"1072650\",\"henryshtang\",\"Henry Tang\",\"05/13/2017\",\"1\"\n\"4671848\",\"mengtongxu\",\"Mengtong XU\",\"03/15/2020\",\"0\"\n\"4672134\",\"xianganqi\",\"XIANG ANQI\",\"03/15/2020\",\"0\"\n\"5345886\",\"snowmeli\",\"Snowmeli\",\"06/22/2020\",\"0\"\n\"5129046\",\"a17186819035\",\"машинное обучение\",\"05/20/2020\",\"0\"\n\"5345891\",\"anuhw0228\",\"Hubert He\",\"06/22/2020\",\"1\"\n\"5345986\",\"josephliweihao\",\"WeiHao Li\",\"06/22/2020\",\"0\"\n\"2137547\",\"gawaingan\",\"Gawain Gan\",\"08/08/2018\",\"0\"\n\"2379406\",\"watmics\",\"Watmics\",\"10/18/2018\",\"1\"\n\"3648601\",\"godxie\",\"JIAQING XIE\",\"08/31/2019\",\"1\"\n\"4600383\",\"sifanwang96\",\"Sifan wang96\",\"03/03/2020\",\"0\"\n\"5067153\",\"guoweiw9\",\"gw.w9\",\"05/11/2020\",\"0\"\n\"4672655\",\"zhaoyujie123\",\"ZhaoYuJie 123\",\"03/15/2020\",\"0\"\n\"3422743\",\"tangyaxu\",\"Patrick\",\"07/03/2019\",\"1\"\n\"4600143\",\"dushaohui\",\"Du Shaohui\",\"03/03/2020\",\"0\"\n\"4711749\",\"xinjianli\",\"Xinjian\",\"03/21/2020\",\"1\"\n\"5202248\",\"oneyiii\",\"noone\",\"05/31/2020\",\"0\"\nHere's a simple/ugly code snippet to get an already ugly visualization in python with networkx:
\nimport numpy as np\nimport pandas as pd\nimport networkx as nx\nimport matplotlib.pyplot as plt\n\nusers = pd.read_csv('../input/meta-kaggle/Users.csv')\nteams = pd.read_csv('../input/meta-kaggle/Teams.csv')\nteammembers = pd.read_csv('../input/meta-kaggle/TeamMemberships.csv')\n\nROOT = <ENTER USER ID HERE>\nnodes = {ROOT}\ndone = set()\n\ngraph = nx.DiGraph()\nfor _ in range(DEPTH): # How far/hops do you want to go from this node?\n to_visit = list(nodes)\n for node in to_visit:\n print(node)\n done.add(node)\n\n teams = teammembers[teammembers['UserId'] == node]['TeamId'].values\n for team in teams:\n for user in teammembers[teammembers['TeamId'] == team]['UserId']:\n if user not in done:\n nodes.add(user)\n\n graph.add_edge(node, user)\n\n nodes.remove(node)\n\npos = nx.drawing.layout.random_layout(graph)\nplt.figure(figsize=(15, 15))\nnx.drawing.nx_pylab.draw(graph, pos=pos)\nnx.drawing.nx_pylab.draw_networkx_labels(graph, pos=pos)\nplt.show()\nAll those customers of Kaggle solutions should be banned FOREVER, shame on them.
", "votes": 9, "replies": [] }, { "id": 972293, "author_name": "Gunes Evitan", "author_url": "", "post_date": "2020-08-16T12:40:03.923000", "content": "Kaggle medal market strikes again
", "votes": 10, "replies": [] }, { "id": 971542, "author_name": "Carl McBride Ellis", "author_url": "", "post_date": "2020-08-15T16:46:13.797000", "content": "I must say that, after learning about the medal farm concept below, if any of the 'Magnificent 7' accounts survive scrutiny by the kaggle team, and also come out of the shakeup having obtained medals, it would be interesting to see the associated user activity over the next few months, for example, changes in Display Name, participation in other competitions, etc. i.e. activity indicative of a change in account ownership.
", "votes": 10, "replies": [ { "id": 971672, "author_name": "Nischay Dhankhar", "author_url": "", "post_date": "2020-08-15T19:43:12.317000", "content": "
Hi all! Thanks for the info. We try to catch these but require conclusive evidence. Please submit information to https://www.kaggle.com/compliance if there is an undetected offense we should investigate.
", "votes": 10, "replies": [ { "id": 970560, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-14T15:02:40.423000", "content": "Sometimes it's (nearly) impossible to get evidence, just by adding a bit bias to a high scoring submission, and submit it using a totally different ip address, you'll never know it. But it doesn't mean that it doesn't exist, in this post all cheats are quite obvious, but some less obvious ones do exist in this competition
", "votes": 4, "replies": [] }, { "id": 971834, "author_name": "Jacek Poplawski", "author_url": "", "post_date": "2020-08-16T01:08:51.773000", "content": "maybe it's time to require real user data when creating Kaggle account…? this way different IP won't help you hiding
", "votes": 2, "replies": [] }, { "id": 974143, "author_name": "Pascal Pfeiffer", "author_url": "", "post_date": "2020-08-17T18:40:52.327000", "content": "What kind of real user data would you suggest?
\nI think it would be very difficult to spot fake IDs given the large number of nationalities competing here.
https://www.kaggle.com/jmayank23
\nhttps://www.kaggle.com/mayankjain23
Wow what a coincidence! Same username, same everything, one placing 8th and the other 13th!
\nhttps://www.kaggle.com/ajibeshmohapatra
\nhttps://www.kaggle.com/jubunmohapatra
And both set to place around the same rank in Wheat as well!
\nMaybe I should change my name, lol
", "votes": 3, "replies": [] }, { "id": 974638, "author_name": "Hitesh Gorana", "author_url": "", "post_date": "2020-08-18T02:09:15.907000", "content": "look at the dataset names😂
\n8 place created fold 1 global wheat
\n13 place created fold 2 global wheat
SAME TIME CREATED ACCOUNTS 2 months ago
\ntaken part in SAME competition
\njust looking at the dataset namefold 2 global wheaton 1 acount andfold 1 global wheatin second account
\nsame name too
This is a smart strategy, so when they delete the one account, he will still have the other 😉
\n
name literally says “#2”
", "votes": 0, "replies": [] } ] } ] }, { "id": 972959, "author_name": "ZachhhBee", "author_url": "", "post_date": "2020-08-17T02:04:22.523000", "content": "The fourth guy in current public LB named 'zhaogongzuo2' literally means 'to find a job' in Chinese pingyin. Why not if you just need to suddenly submit a great score in the final week and then get yourself one?
\nCertain group of people are becoming fearless on kaggle.
", "votes": 7, "replies": [ { "id": 973321, "author_name": "StepD", "author_url": "", "post_date": "2020-08-17T08:41:33.713000", "content": "🤣 They are crazy
", "votes": 0, "replies": [] } ] }, { "id": 972008, "author_name": "Aziz_Belaweid", "author_url": "", "post_date": "2020-08-16T06:48:52.250000", "content": "someone just got 0.9655 using one submission, this is frustrating my team and I have been working for over a month to get to 0.9649 and bam just like that a novice gets higher ranking than you.
", "votes": 7, "replies": [ { "id": 972078, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2020-08-16T08:01:08.920000", "content": "I think someone shared a notebook that scores ~0.965 just recently. (which should not have happened since it's less than 7 days until private LB reveal)
", "votes": 2, "replies": [] }, { "id": 972080, "author_name": "", "author_url": "", "post_date": "2020-08-16T08:01:16.023000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 971888, "author_name": "ld", "author_url": "", "post_date": "2020-08-16T03:46:36.997000", "content": "Kaggle will remove them since they are novices. They would be allowed to cheat if they were Kaggle masters and grandmasters.
\nhttps://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard
\nhttps://www.kaggle.com/c/imet-2019-fgvc6/discussion/94430
This is bigger than you think. I was manually pulling up accounts from that site as fast as I could, and many are \"sleeper\" accounts. In Kaggle and have not even entered any comps yet. It's a mess. All those people cheated in Whale, DeepFake, M5, Pancreatic, now this comp, and so many more. People that would have gotten a medal, or a better medal…….cheated, so someone could make money.
", "votes": 3, "replies": [] }, { "id": 973200, "author_name": "sin", "author_url": "", "post_date": "2020-08-17T06:59:32.623000", "content": "Ofc it’s ValeriyBabushkin from ods.ai😄
", "votes": 7, "replies": [] }, { "id": 973244, "author_name": "Dieter", "author_url": "", "post_date": "2020-08-17T07:36:39.773000", "content": "https://www.kaggle.com/c/landmark-recognition-2019/leaderboard
\n
I think @lifeifei38 did us a huge favor explaining that what they do is sharing public notebooks that are available for their students but not to the rest of the competitors.
\nThis argument is more than enough for kaggle to take actions because we know that in this case, kaggle algorithm won't detect any fraud, but the teacher himself explained the situation.
\nLet's believe that it's just a normal course explaining some deep learning basics and the students are smart enough to make their way to the medal zone with their own work and effort, they still worked with their course's pipelines that are not available for the rest of the competitors.
\nThanks @group16 for your investigation :)
\n", "votes": 5, "replies": [ { "id": 973538, "author_name": "FChmiel", "author_url": "", "post_date": "2020-08-17T10:57:27.217000", "content": "I'm highly skeptical that all those students 'are smart enough to make their way to the medal zone with their own work and effort'. Evidence suggest otherwise.
", "votes": 0, "replies": [] } ] }, { "id": 973481, "author_name": "Mukharbek Organokov", "author_url": "", "post_date": "2020-08-17T10:21:40.273000", "content": "@group16 My personal respect man! Really great job.
", "votes": 5, "replies": [] }, { "id": 973265, "author_name": "Changyi", "author_url": "", "post_date": "2020-08-17T07:49:16.273000", "content": "So scaring, they're like bots… Even now they continue to submit !
", "votes": 5, "replies": [] }, { "id": 973194, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T06:55:08.753000", "content": "This has been going on a LONG time.
\nThey just took the site down. Site had MANY kaggle comps, with students registered 20, 65, even over a hundred students in the Whale Tail comp.
\nStudents are all BRAND NEW kaggle accounts. No activity, no comments, no posts. They team with other students, typically 4-5 in a team. Everyone has the same deal, no comments, no activity, no posts. They ALWAYS place toward the top of the comp, either medaling or very close to it.
\nKaggle is infiltrated, this is a real mess.
", "votes": 5, "replies": [] }, { "id": 973213, "author_name": "Firat Gonen", "author_url": "", "post_date": "2020-08-17T07:10:22.270000", "content": "so many rules broken or bent
", "votes": 5, "replies": [] }, { "id": 975009, "author_name": "Stelios S", "author_url": "", "post_date": "2020-08-18T05:43:08.400000", "content": "In the mind of a buyer of this kind of \"services\":
\nRecruiter: So,what you have done so far?
\nBought code/acc user : Gold medal on Kaggle in Public Leaderboard
\nRecruiter: You're hired.
\n==============================
\nRecruiter: So,what you have done so far?
\nBought code/acc user : Gold medal on Kaggle in Public Leaderboard.
\nRecruiter : We will let you know (Spoiler alert :They won't let you know, in most of the cases ).
\nJob candidate closes door
\nRecruiter : Surprise
", "votes": 3, "replies": [] }, { "id": 974154, "author_name": "Prateek Mishra", "author_url": "", "post_date": "2020-08-17T18:52:10.453000", "content": "There are so many fake accounts holding high ranks on public LB.
\nThis team of 5 members with 4 novices is cheating too. All 4 of the novices didn't even make a single submission.
\n
\n@group16 you are right they are surely going to sell these accounts because I don't find any good reason behind doing this.
\nNone of the accounts were active in any past competitions. They are only logged in for participating in team.
\n[P.S. There are many such accounts.]
\n@juliaelliott @kaggle do something. They are stealing medals from deserved ones.
Please rest assured, we will be taking action.
", "votes": 26, "replies": [] }, { "id": 974176, "author_name": "Prateek Mishra", "author_url": "", "post_date": "2020-08-17T19:12:31.580000", "content": "Thanks…!
", "votes": 1, "replies": [] }, { "id": 974191, "author_name": "yimacs", "author_url": "", "post_date": "2020-08-17T19:29:55.693000", "content": "@inversion thanks for the confirmation.
", "votes": 1, "replies": [] }, { "id": 974301, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T21:21:33.973000", "content": "@prateek0x the yikououbao team is particularly brazen, they have taken many medals and stand to do so in two more comps in the next few hours. They have no shame.
", "votes": 1, "replies": [] }, { "id": 974319, "author_name": "Jie Lu", "author_url": "", "post_date": "2020-08-17T22:10:19.857000", "content": "@inversion you should also check for wheat detection and landmark retrieval 2020, they will also terminate soon
", "votes": 2, "replies": [] }, { "id": 974400, "author_name": "inversion", "author_url": "", "post_date": "2020-08-18T00:11:06.220000", "content": "We are checking well beyond just this competition. :-)
", "votes": 14, "replies": [] }, { "id": 976895, "author_name": "Signal", "author_url": "", "post_date": "2020-08-19T06:53:36.640000", "content": "I'll be really disappointed if the yikououbao isn't disqualified, look at the medals they have taken. They took in Melanoma and now are set for gold in Wheat. Seriously, the most brazen of groups.
", "votes": 2, "replies": [] }, { "id": 977422, "author_name": "YaGana Sheriff-Hussaini", "author_url": "", "post_date": "2020-08-19T13:17:28.057000", "content": "Thanks @inversion. The yikououbao account does look fishy. Always teaming up with noobs and only one repeated team member (i.e. yauwaiyim) etc. and there is a deleted account in the ALASKA2 competition among his/her team members.
", "votes": 0, "replies": [] } ] }, { "id": 973444, "author_name": "Konstantin Yakovlev", "author_url": "", "post_date": "2020-08-17T09:58:00.747000", "content": "I think they have a bunch of accounts for public leaderboard probing. What for? - It's still a mystery for me but I think that they will not score well on private part.
\nAny \"normal\" account will not have enough \"resources\" to conduct more than 5-10 experiments per day. But the number of accounts involved shows that they are doing 50-100 probes daily. With constant daily LB progress it could be only public probing.
\nThere is a \"question\" in my theory. Why they are not doing single record probing with low score and not using one or two accounts for \"public lb promotion sell\"?
", "votes": 3, "replies": [ { "id": 973459, "author_name": "Signal", "author_url": "", "post_date": "2020-08-17T10:05:02.027000", "content": "They had a few years worth of comps on their website. Years. Each comp had 25-100+ \"students\" registered in it. The operation is bigger than we know.
", "votes": 1, "replies": [] }, { "id": 973464, "author_name": "Changyi", "author_url": "", "post_date": "2020-08-17T10:09:41.263000", "content": "Let's monitor the private LB. if a team get 0.98 or 0.99 in private without sharing their method I won't be convinced.
", "votes": 1, "replies": [] }, { "id": 973484, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2020-08-17T10:23:20.263000", "content": "It gave them a short window to advertise to gullible people not knowing that there is a difference between public and private LB.
\n\"BUY A THIRD PLACE ON KAGGLE COMPETITION WITH OVER 3000 CONTESTANTS NOW FOR ONLY 500 DOLLARS\"!
", "votes": 3, "replies": [] } ] }, { "id": 973429, "author_name": "Md Fahim", "author_url": "", "post_date": "2020-08-17T09:46:54.733000", "content": "One of them in the money zone in this competition. He may recover his payment. More surprisingly this is his 1st competition.
", "votes": 3, "replies": [ { "id": 973442, "author_name": "Changyi", "author_url": "", "post_date": "2020-08-17T09:57:00.700000", "content": "No need to worry about the top ones. I believe that they'll be deleted.
\nI worry more about the ones hide in silver/bronze whom are less obvious. They take their first medal in this competition and next time they are no more novices so people won't doubt them.