![\"Doge](\"https://i.postimg.cc/2S59TB5n/if-doge.png\")
It saddens my heart to see some people try to increase the number of their daily submissions so inelegantly, in the future please put you heart and soul into it, like using more different breeds of dogs as profile pics (cute pictures BTW, I love them! specially the Shiba Inu picture) or not using different combinations of w and wanliyu for generating usernames.
", "messages": [ { "id": 537628, "postDate": "2019-05-27T11:10:38.717Z", "content": "It saddens my heart to see some people try to increase the number of their daily submissions so inelegantly, in the future please put you heart and soul into it, like using more different breeds of dogs as profile pics (cute pictures BTW, I love them! specially the Shiba Inu picture) or not using different combinations of w and wanliyu for generating usernames.
", "rawMarkdown": "It saddens my heart to see some people try to increase the number of their daily submissions so inelegantly, in the future please put you heart and soul into it, like using more different breeds of dogs as profile pics (cute pictures BTW, I love them! specially the Shiba Inu picture) or not using different combinations of **w** and **wanliyu** for generating usernames.\n\n", "votes": 20 }, { "id": 537733, "postDate": "2019-05-27T14:20:14.883Z", "content": "those who has more w also had exact same score (1.288) before team merger...hmm... \nis it possible without any private sharing outside teams ?
", "rawMarkdown": "those who has more **w** also had exact same score (1.288) before team merger...hmm... \nis it possible without any **private sharing outside teams** ? ", "votes": 5, "replies": [ { "id": 537746, "postDate": "2019-05-27T14:51:26.010Z", "content": "They are in a team now, aren't they? What's the issue then?
", "rawMarkdown": "They are in a team now, aren't they? What's the issue then?", "votes": 4 }, { "id": 537783, "postDate": "2019-05-27T15:43:34.143Z", "content": "LOL, I never get why asking a question can lead to downvote. This is fascinating.
", "rawMarkdown": "LOL, I never get why asking a question can lead to downvote. This is fascinating.", "votes": 5 }, { "id": 537805, "postDate": "2019-05-27T16:09:42.107Z", "content": "i might misunderstood the rules. I thought private sharing was strictly forbidden before merger, even if you team up later on. \nIMHO even if it is not a solid violation, I think it is not fair.
\n\nps. @CPMP i dont know who (or why) downvote your post. it is not me :)
", "rawMarkdown": "i might misunderstood the rules. I thought private sharing was strictly forbidden before merger, even if you team up later on. \nIMHO even if it is not a solid violation, I think it is not fair.\n\nps. @CPMP i dont know who (or why) downvote your post. it is not me :)", "votes": 1 }, { "id": 537810, "postDate": "2019-05-27T16:21:35.483Z", "content": "How would you prove some sharing happened before merger?
", "rawMarkdown": "How would you prove some sharing happened before merger?", "votes": 1 }, { "id": 537815, "postDate": "2019-05-27T16:29:35.457Z", "content": "To be honest I also understand the rules this way - you can't share the code even if you finally merge. \nKaggle could find it if someone is using the same code in kernels or submit the same submissions. No matter Kaggle will do it, it's against the rules
", "rawMarkdown": "To be honest I also understand the rules this way - you can't share the code even if you finally merge. \nKaggle could find it if someone is using the same code in kernels or submit the same submissions. No matter Kaggle will do it, it's against the rules", "votes": 1 }, { "id": 538097, "postDate": "2019-05-28T06:05:21Z", "content": "This may be the subject of another competition. detecting private sharing before team merger.
", "rawMarkdown": "This may be the subject of another competition. detecting private sharing before team merger. ", "votes": 1 }, { "id": 538340, "postDate": "2019-05-28T12:47:20.843Z", "content": "\n\n\nyou can't share the code even if you finally merge.
\n
You can't share anything outside teams, why focus on code only?
", "rawMarkdown": "> you can't share the code even if you finally merge. \n\nYou can't share anything outside teams, why focus on code only?" }, { "id": 538353, "postDate": "2019-05-28T13:03:26.470Z", "content": "I think thats bit grey area. i am not quite sure how will you team up with someone unless you exchange something to be sure.
", "rawMarkdown": "I think thats bit grey area. i am not quite sure how will you team up with someone unless you exchange something to be sure.", "votes": 2 } ] }, { "id": 537655, "postDate": "2019-05-27T12:01:43.307Z", "content": "These are epic usernames\n1. wanliyu\n2. wwwanliyu\n3. wwwwanliyu
", "rawMarkdown": "These are epic usernames\n1. wanliyu\n2. wwwanliyu\n3. wwwwanliyu\n\n", "votes": 3, "replies": [ { "id": 537897, "postDate": "2019-05-27T19:58:13.387Z", "content": "what are the rankings ?
", "rawMarkdown": "what are the rankings ?" }, { "id": 537898, "postDate": "2019-05-27T19:58:35.710Z", "content": "never mind
", "rawMarkdown": "never mind" } ] }, { "id": 540282, "postDate": "2019-05-31T09:24:54.470Z", "content": "
The Good, the Bad and the Ugly ...!
\n\nI don't want to judge any non-technical efforts in this competition, but cheating should be bold in kaggle such a way that we have some Anti-Cheating algorithms to detect cheaters (fraud detection). So the topic of following link should be useful to look at this repulsive behavior in a different view:
\n\n\"How Winning Leads to Cheating?!\"
\n\n\"Perhaps competitions have long-lasting effects....”
", "rawMarkdown": "The Good, the Bad and the Ugly ...!\n\nI don't want to judge any non-technical efforts in this competition, but cheating should be bold in kaggle such a way that we have some Anti-Cheating algorithms to detect cheaters (fraud detection). So the topic of following link should be useful to look at this repulsive behavior in a different view:\n\n[\"How Winning Leads to Cheating?!\"](https://www.scientificamerican.com/article/how-winning-leads-to-cheating/?redirect=1) \n\n\n\"Perhaps competitions have long-lasting effects....”\n", "votes": 1 }, { "id": 537772, "postDate": "2019-05-27T15:29:37.587Z", "content": "Autumn?
", "rawMarkdown": "Autumn?", "votes": 1, "replies": [ { "id": 537781, "postDate": "2019-05-27T15:43:10.477Z", "content": "Can't be as he told us he is definitely not a cheater - the fact his scores were deleted from Santander and one of the accounts was deleted must be just a coincidence
", "rawMarkdown": "Can't be as he told us he is definitely not a cheater - the fact his scores were deleted from Santander and one of the accounts was deleted must be just a coincidence", "votes": 9 }, { "id": 538368, "postDate": "2019-05-28T13:31:08.163Z", "content": "Here is the \"Autumn\" thread in case people don't get the reference:\nhttps://www.kaggle.com/c/santander-customer-transaction-prediction/discussion/83786#latest-517964
", "rawMarkdown": "Here is the \"Autumn\" thread in case people don't get the reference:\nhttps://www.kaggle.com/c/santander-customer-transaction-prediction/discussion/83786#latest-517964", "votes": 2 } ] }, { "id": 537758, "postDate": "2019-05-27T15:09:55.427Z", "content": "I think it is trouble even for human to distinguish between serious and sarcasm/humor at times. Anytime you accuse people of cheating, you run the risk of a false positive for your own reputation and the accused party(s).
\n\nNevertheless, how does kaggle deal with fictitious accounts that were used to submit a couple of extra entries for a competition nearing completion, accounts which were combined to the main account/team so as to not cause suspicion? For ex: if you have submitted below a theoretical max of 300 team submissions, and there is only 10 days left. If you have 1 accounts, then you can only submit 20 max entries. If you create 2 more, you can triple your submissions, and still combine teams with fictitious accounts so long as you're under the theoretical limit. How would kaggle deal with such a loophope?
", "rawMarkdown": "I think it is trouble even for human to distinguish between serious and sarcasm/humor at times. Anytime you accuse people of cheating, you run the risk of a false positive for your own reputation and the accused party(s).\n\nNevertheless, how does kaggle deal with fictitious accounts that were used to submit a couple of extra entries for a competition nearing completion, accounts which were combined to the main account/team so as to not cause suspicion? For ex: if you have submitted below a theoretical max of 300 team submissions, and there is only 10 days left. If you have 1 accounts, then you can only submit 20 max entries. If you create 2 more, you can triple your submissions, and still combine teams with fictitious accounts so long as you're under the theoretical limit. How would kaggle deal with such a loophope?", "votes": 1, "replies": [ { "id": 537770, "postDate": "2019-05-27T15:26:36.083Z", "content": "This happened in Santander. There were 3 solo people, who mentioned on the forums that they were all working together. Then they merged halfway through the competition (by force). It is shady, because you can game the maximum submission like you said.
", "rawMarkdown": "This happened in Santander. There were 3 solo people, who mentioned on the forums that they were all working together. Then they merged halfway through the competition (by force). It is shady, because you can game the maximum submission like you said.", "votes": 1 }, { "id": 537780, "postDate": "2019-05-27T15:41:15.513Z", "content": "We remove cheaters at the end of the competition. Note that participants who merge teams must still have less than the total number of submissions before they are allowed to merge. Therefore if all team members use all their daily submissions, and try to merge at the end, the system won't allow it as they have more than the total allowable submissions.
", "rawMarkdown": "We remove cheaters at the end of the competition. Note that participants who merge teams must still have less than the total number of submissions before they are allowed to merge. Therefore if all team members use all their daily submissions, and try to merge at the end, the system won't allow it as they have more than the total allowable submissions.", "votes": 2 }, { "id": 538055, "postDate": "2019-05-28T04:00:30.543Z", "content": "Just a thought- but what about someone who joined the competition late? If they had joined 7 days prior to the deadline they would only have 14 submissions. But with 5 accounts they get 14 * 5 while remaining below the total submission threshold. More submissions later in the competition are more valuable since they get the benefit of using the submissions after much of the public breakthroughs/kernels have been shared.
\n\nThis of course only matters in the instance of one person with multiple accounts. Which I’m not saying is the case for those being accused in this thread. For all I know they are totally legit and all love puppies! 🐶
", "rawMarkdown": "Just a thought- but what about someone who joined the competition late? If they had joined 7 days prior to the deadline they would only have 14 submissions. But with 5 accounts they get 14 * 5 while remaining below the total submission threshold. More submissions later in the competition are more valuable since they get the benefit of using the submissions after much of the public breakthroughs/kernels have been shared.\n\nThis of course only matters in the instance of one person with multiple accounts. Which I’m not saying is the case for those being accused in this thread. For all I know they are totally legit and all love puppies! 🐶", "votes": 2 } ] }, { "id": 537634, "postDate": "2019-05-27T11:21:50.493Z", "content": "I wonder how cheaters are detected. Should be a fun but difficult problem for kaggle to handle.
", "rawMarkdown": "I wonder how cheaters are detected. Should be a fun but difficult problem for kaggle to handle.", "votes": 1 }, { "id": 538337, "postDate": "2019-05-28T12:41:10.650Z", "rawMarkdown": "", "isDeleted": true }, { "id": 538321, "postDate": "2019-05-28T12:21:03.593Z", "rawMarkdown": "", "votes": -2, "isDeleted": true }, { "id": 537684, "postDate": "2019-05-27T13:08:40.997Z", "rawMarkdown": "", "votes": -4, "isDeleted": true, "replies": [ { "id": 537699, "postDate": "2019-05-27T13:27:44.350Z", "content": "Are you saying the accusation is a bit woof?
\n\nSeriously though, I hope you aren't cheating as your great score on Don't Overfit competition will probably be looked at too!
", "rawMarkdown": "Are you saying the accusation is a bit woof?\n\nSeriously though, I hope you aren't cheating as your great score on Don't Overfit competition will probably be looked at too!", "votes": 7 }, { "id": 537777, "postDate": "2019-05-27T15:39:44.623Z", "content": "Wanyliu - It sounds like you're admitting to violating the competition rules. The official rules state that you are not allowed to share code or collaborate outside of Kaggle teams. In order to not be disqualified by the end of the competition - you need to merge into a single team.
", "rawMarkdown": "Wanyliu - It sounds like you're admitting to violating the competition rules. The official rules state that you are not allowed to share code or collaborate outside of Kaggle teams. In order to not be disqualified by the end of the competition - you need to merge into a single team.", "votes": 2 }, { "id": 537787, "postDate": "2019-05-27T15:45:34.997Z", "content": "I don't think he/she admitted cheating, to the contrary. I frankly find that this accusation based on avatars is strange.
", "rawMarkdown": "I don't think he/she admitted cheating, to the contrary. I frankly find that this accusation based on avatars is strange. " }, { "id": 537800, "postDate": "2019-05-27T16:00:46.857Z", "content": "the question is are these four participant are single person ? if true then rule is violated
", "rawMarkdown": "the question is are these four participant are single person ? if true then rule is violated", "votes": 2 }, { "id": 537811, "postDate": "2019-05-27T16:23:02.470Z", "content": "That's a (fair) question. It is different from accusing upfront ...
", "rawMarkdown": "That's a (fair) question. It is different from accusing upfront ...", "votes": 2 }, { "id": 537816, "postDate": "2019-05-27T16:31:21.877Z", "content": "@cpmpml I think it's also based on the names. Even if it's a false accusation, it still funny that Kain pointed it out, the quirk.
\n\nTo clarify on my comment - Wanliyu said \"We have four participants\" - which I thought implied a team, hence my comment. Once the competition has concluded, we will run our anti-cheating algorithm to remove any potential cheaters. We do not disclose the details of this algorithm.
", "rawMarkdown": "To clarify on my comment - Wanliyu said \"We have four participants\" - which I thought implied a team, hence my comment. Once the competition has concluded, we will run our anti-cheating algorithm to remove any potential cheaters. We do not disclose the details of this algorithm.", "votes": 4 }, { "id": 537864, "postDate": "2019-05-27T19:05:34.717Z", "content": "please don't disclose. people are pretty smart here, they will start using the knowledge in their benefit
", "rawMarkdown": "please don't disclose. people are pretty smart here, they will start using the knowledge in their benefit", "votes": 2 }, { "id": 537891, "postDate": "2019-05-27T19:44:51.797Z", "content": "Does Kaggle have an appeal process for this algorithm? No algorithm is perfect, and surely some legitimate competitors would fall foul of it.
", "rawMarkdown": "Does Kaggle have an appeal process for this algorithm? No algorithm is perfect, and surely some legitimate competitors would fall foul of it.", "votes": 2 }, { "id": 538225, "postDate": "2019-05-28T10:13:23.763Z", "content": "@addisonhoward A \"Kaggle cheater detection\" competition would be awesome.
", "rawMarkdown": "@addisonhoward A \"Kaggle cheater detection\" competition would be awesome.", "votes": 1 }, { "id": 538230, "postDate": "2019-05-28T10:16:24.567Z", "content": "\n\n\nwhich I thought implied a team
\n
Well, he is in a team of 4, which is what I think he wrote.
", "rawMarkdown": "> which I thought implied a team\n\nWell, he is in a team of 4, which is what I think he wrote." }, { "id": 538239, "postDate": "2019-05-28T10:20:24.023Z", "content": "All I am saying is that saying they are cheaters based on avatars or names isn't right.
\n\nAnd when we're at it, I suggest that you look for kagglers that always teamed with the same person and never entered a competition alone. I would not be surprised if these are fake accounts.
\n\nBut, given they team, I am not sure there is a way to prove any rule violation.
", "rawMarkdown": "All I am saying is that saying they are cheaters based on avatars or names isn't right. \n\nAnd when we're at it, I suggest that you look for kagglers that always teamed with the same person and never entered a competition alone. I would not be surprised if these are fake accounts. \n\nBut, given they team, I am not sure there is a way to prove any rule violation.", "votes": 1 }, { "id": 538317, "postDate": "2019-05-28T12:09:44.743Z", "content": "@pestipeti I think the problem is similar to fraud modeling. Imbalanced dataset. However, I'm guessing they're probably rules based at the moment.
", "rawMarkdown": "@pestipeti I think the problem is similar to fraud modeling. Imbalanced dataset. However, I'm guessing they're probably rules based at the moment." } ] } ], "comments": [ { "id": 537733, "author_name": "Nefsi", "author_url": "", "post_date": "2019-05-27T14:20:14.883000", "content": "those who has more w also had exact same score (1.288) before team merger...hmm... \nis it possible without any private sharing outside teams ?
", "votes": 5, "replies": [ { "id": 537746, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-27T14:51:26.010000", "content": "They are in a team now, aren't they? What's the issue then?
", "votes": 4, "replies": [] }, { "id": 537783, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-27T15:43:34.143000", "content": "LOL, I never get why asking a question can lead to downvote. This is fascinating.
", "votes": 5, "replies": [] }, { "id": 537805, "author_name": "Nefsi", "author_url": "", "post_date": "2019-05-27T16:09:42.107000", "content": "i might misunderstood the rules. I thought private sharing was strictly forbidden before merger, even if you team up later on. \nIMHO even if it is not a solid violation, I think it is not fair.
\n\nps. @CPMP i dont know who (or why) downvote your post. it is not me :)
", "votes": 1, "replies": [] }, { "id": 537810, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-27T16:21:35.483000", "content": "How would you prove some sharing happened before merger?
", "votes": 1, "replies": [] }, { "id": 537815, "author_name": "DavidS", "author_url": "", "post_date": "2019-05-27T16:29:35.457000", "content": "To be honest I also understand the rules this way - you can't share the code even if you finally merge. \nKaggle could find it if someone is using the same code in kernels or submit the same submissions. No matter Kaggle will do it, it's against the rules
", "votes": 1, "replies": [] }, { "id": 538097, "author_name": "Nefsi", "author_url": "", "post_date": "2019-05-28T06:05:21", "content": "This may be the subject of another competition. detecting private sharing before team merger.
", "votes": 1, "replies": [] }, { "id": 538340, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-28T12:47:20.843000", "content": "\n\n\nyou can't share the code even if you finally merge.
\n
You can't share anything outside teams, why focus on code only?
", "votes": 0, "replies": [] }, { "id": 538353, "author_name": "Manoj", "author_url": "", "post_date": "2019-05-28T13:03:26.470000", "content": "I think thats bit grey area. i am not quite sure how will you team up with someone unless you exchange something to be sure.
", "votes": 2, "replies": [] } ] }, { "id": 537655, "author_name": "Manoj", "author_url": "", "post_date": "2019-05-27T12:01:43.307000", "content": "These are epic usernames\n1. wanliyu\n2. wwwanliyu\n3. wwwwanliyu
", "votes": 3, "replies": [ { "id": 537897, "author_name": "Firat Gonen", "author_url": "", "post_date": "2019-05-27T19:58:13.387000", "content": "what are the rankings ?
", "votes": 0, "replies": [] }, { "id": 537898, "author_name": "Firat Gonen", "author_url": "", "post_date": "2019-05-27T19:58:35.710000", "content": "never mind
", "votes": 0, "replies": [] } ] }, { "id": 540282, "author_name": "Hassan Abedi", "author_url": "", "post_date": "2019-05-31T09:24:54.470000", "content": "
The Good, the Bad and the Ugly ...!
\n\nI don't want to judge any non-technical efforts in this competition, but cheating should be bold in kaggle such a way that we have some Anti-Cheating algorithms to detect cheaters (fraud detection). So the topic of following link should be useful to look at this repulsive behavior in a different view:
\n\n\"How Winning Leads to Cheating?!\"
\n\n\"Perhaps competitions have long-lasting effects....”
", "votes": 1, "replies": [] }, { "id": 537772, "author_name": "interneuron", "author_url": "", "post_date": "2019-05-27T15:29:37.587000", "content": "Autumn?
", "votes": 1, "replies": [ { "id": 537781, "author_name": "Scirpus", "author_url": "", "post_date": "2019-05-27T15:43:10.477000", "content": "Can't be as he told us he is definitely not a cheater - the fact his scores were deleted from Santander and one of the accounts was deleted must be just a coincidence
", "votes": 9, "replies": [] }, { "id": 538368, "author_name": "Rob Mulla", "author_url": "", "post_date": "2019-05-28T13:31:08.163000", "content": "Here is the \"Autumn\" thread in case people don't get the reference:\nhttps://www.kaggle.com/c/santander-customer-transaction-prediction/discussion/83786#latest-517964
", "votes": 2, "replies": [] } ] }, { "id": 537758, "author_name": "Tim Yee", "author_url": "", "post_date": "2019-05-27T15:09:55.427000", "content": "I think it is trouble even for human to distinguish between serious and sarcasm/humor at times. Anytime you accuse people of cheating, you run the risk of a false positive for your own reputation and the accused party(s).
\n\nNevertheless, how does kaggle deal with fictitious accounts that were used to submit a couple of extra entries for a competition nearing completion, accounts which were combined to the main account/team so as to not cause suspicion? For ex: if you have submitted below a theoretical max of 300 team submissions, and there is only 10 days left. If you have 1 accounts, then you can only submit 20 max entries. If you create 2 more, you can triple your submissions, and still combine teams with fictitious accounts so long as you're under the theoretical limit. How would kaggle deal with such a loophope?
", "votes": 1, "replies": [ { "id": 537770, "author_name": "CoreyJamesLevinson", "author_url": "", "post_date": "2019-05-27T15:26:36.083000", "content": "This happened in Santander. There were 3 solo people, who mentioned on the forums that they were all working together. Then they merged halfway through the competition (by force). It is shady, because you can game the maximum submission like you said.
", "votes": 1, "replies": [] }, { "id": 537780, "author_name": "Addison Howard", "author_url": "", "post_date": "2019-05-27T15:41:15.513000", "content": "We remove cheaters at the end of the competition. Note that participants who merge teams must still have less than the total number of submissions before they are allowed to merge. Therefore if all team members use all their daily submissions, and try to merge at the end, the system won't allow it as they have more than the total allowable submissions.
", "votes": 2, "replies": [] }, { "id": 538055, "author_name": "Rob Mulla", "author_url": "", "post_date": "2019-05-28T04:00:30.543000", "content": "Just a thought- but what about someone who joined the competition late? If they had joined 7 days prior to the deadline they would only have 14 submissions. But with 5 accounts they get 14 * 5 while remaining below the total submission threshold. More submissions later in the competition are more valuable since they get the benefit of using the submissions after much of the public breakthroughs/kernels have been shared.
\n\nThis of course only matters in the instance of one person with multiple accounts. Which I’m not saying is the case for those being accused in this thread. For all I know they are totally legit and all love puppies! 🐶
", "votes": 2, "replies": [] } ] }, { "id": 537634, "author_name": "Tim Yee", "author_url": "", "post_date": "2019-05-27T11:21:50.493000", "content": "I wonder how cheaters are detected. Should be a fun but difficult problem for kaggle to handle.
", "votes": 1, "replies": [] }, { "id": 538337, "author_name": "", "author_url": "", "post_date": "2019-05-28T12:41:10.650000", "content": "", "votes": 0, "replies": [] }, { "id": 538321, "author_name": "", "author_url": "", "post_date": "2019-05-28T12:21:03.593000", "content": "", "votes": -2, "replies": [] }, { "id": 537684, "author_name": "", "author_url": "", "post_date": "2019-05-27T13:08:40.997000", "content": "", "votes": -4, "replies": [ { "id": 537699, "author_name": "Scirpus", "author_url": "", "post_date": "2019-05-27T13:27:44.350000", "content": "Are you saying the accusation is a bit woof?
\n\nSeriously though, I hope you aren't cheating as your great score on Don't Overfit competition will probably be looked at too!
", "votes": 7, "replies": [] }, { "id": 537777, "author_name": "Addison Howard", "author_url": "", "post_date": "2019-05-27T15:39:44.623000", "content": "Wanyliu - It sounds like you're admitting to violating the competition rules. The official rules state that you are not allowed to share code or collaborate outside of Kaggle teams. In order to not be disqualified by the end of the competition - you need to merge into a single team.
", "votes": 2, "replies": [] }, { "id": 537787, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-27T15:45:34.997000", "content": "I don't think he/she admitted cheating, to the contrary. I frankly find that this accusation based on avatars is strange.
", "votes": 0, "replies": [] }, { "id": 537800, "author_name": "Manoj", "author_url": "", "post_date": "2019-05-27T16:00:46.857000", "content": "the question is are these four participant are single person ? if true then rule is violated
", "votes": 2, "replies": [] }, { "id": 537811, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-27T16:23:02.470000", "content": "That's a (fair) question. It is different from accusing upfront ...
", "votes": 2, "replies": [] }, { "id": 537816, "author_name": "Tim Yee", "author_url": "", "post_date": "2019-05-27T16:31:21.877000", "content": "@cpmpml I think it's also based on the names. Even if it's a false accusation, it still funny that Kain pointed it out, the quirk.
\n\nTo clarify on my comment - Wanliyu said \"We have four participants\" - which I thought implied a team, hence my comment. Once the competition has concluded, we will run our anti-cheating algorithm to remove any potential cheaters. We do not disclose the details of this algorithm.
", "votes": 4, "replies": [] }, { "id": 537864, "author_name": "Manoj", "author_url": "", "post_date": "2019-05-27T19:05:34.717000", "content": "please don't disclose. people are pretty smart here, they will start using the knowledge in their benefit
", "votes": 2, "replies": [] }, { "id": 537891, "author_name": "RNA", "author_url": "", "post_date": "2019-05-27T19:44:51.797000", "content": "Does Kaggle have an appeal process for this algorithm? No algorithm is perfect, and surely some legitimate competitors would fall foul of it.
", "votes": 2, "replies": [] }, { "id": 538225, "author_name": "Peter", "author_url": "", "post_date": "2019-05-28T10:13:23.763000", "content": "@addisonhoward A \"Kaggle cheater detection\" competition would be awesome.
", "votes": 1, "replies": [] }, { "id": 538230, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-28T10:16:24.567000", "content": "\n\n\nwhich I thought implied a team
\n
Well, he is in a team of 4, which is what I think he wrote.
", "votes": 0, "replies": [] }, { "id": 538239, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-28T10:20:24.023000", "content": "All I am saying is that saying they are cheaters based on avatars or names isn't right.
\n\nAnd when we're at it, I suggest that you look for kagglers that always teamed with the same person and never entered a competition alone. I would not be surprised if these are fake accounts.
\n\nBut, given they team, I am not sure there is a way to prove any rule violation.
", "votes": 1, "replies": [] }, { "id": 538317, "author_name": "Tim Yee", "author_url": "", "post_date": "2019-05-28T12:09:44.743000", "content": "@pestipeti I think the problem is similar to fraud modeling. Imbalanced dataset. However, I'm guessing they're probably rules based at the moment.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "537628": "It saddens my heart to see some people try to increase the number of their daily submissions so inelegantly, in the future please put you heart and soul into it, like using more different breeds of dogs as profile pics (cute pictures BTW, I love them! specially the Shiba Inu picture) or not using different combinations of **w** and **wanliyu** for generating usernames.\n\n", "537733": "those who has more **w** also had exact same score (1.288) before team merger...hmm... \nis it possible without any **private sharing outside teams** ? ", "537655": "These are epic usernames\n1. wanliyu\n2. wwwanliyu\n3. wwwwanliyu\n\n", "540282": "", "539312": "The Good, the Bad and the Ugly ...!\n\nI don't want to judge any non-technical efforts in this competition, but cheating should be bold in kaggle such a way that we have some Anti-Cheating algorithms to detect cheaters (fraud detection). So the topic of following link should be useful to look at this repulsive behavior in a different view:\n\n[\"How Winning Leads to Cheating?!\"](https://www.scientificamerican.com/article/how-winning-leads-to-cheating/?redirect=1) \n\n\n\"Perhaps competitions have long-lasting effects....”\n", "537772": "Autumn?", "537758": "I think it is trouble even for human to distinguish between serious and sarcasm/humor at times. Anytime you accuse people of cheating, you run the risk of a false positive for your own reputation and the accused party(s).\n\nNevertheless, how does kaggle deal with fictitious accounts that were used to submit a couple of extra entries for a competition nearing completion, accounts which were combined to the main account/team so as to not cause suspicion? For ex: if you have submitted below a theoretical max of 300 team submissions, and there is only 10 days left. If you have 1 accounts, then you can only submit 20 max entries. If you create 2 more, you can triple your submissions, and still combine teams with fictitious accounts so long as you're under the theoretical limit. How would kaggle deal with such a loophope?", "537634": "I wonder how cheaters are detected. Should be a fun but difficult problem for kaggle to handle.", "538337": "", "538321": "", "537684": "" } }