![\"\"](\"https://i.imgur.com/WW4CFz3.png\")
#11 and #14 are at similar score and submit exactly at the same time. They are from the same country and follow each other. Apologies if it is a coincidence but submitting at the same time is another unlikely thing.
\nWhile they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.
", "messages": [ { "id": 1180301, "postDate": "2021-02-01T07:56:17.063Z", "content": "
#11 and #14 are at similar score and submit exactly at the same time. They are from the same country and follow each other. Apologies if it is a coincidence but submitting at the same time is another unlikely thing.
\nWhile they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.
", "rawMarkdown": "\n\n**#11** and **#14** are at similar score and **submit exactly at the same time**. They are from the same country and follow each other. Apologies if it is a coincidence but submitting at the same time is another unlikely thing.\n\nWhile they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.", "votes": 25 }, { "id": 1180438, "postDate": "2021-02-01T08:56:42.837Z", "content": "same university btw π
", "rawMarkdown": "same university btw π", "votes": 12 }, { "id": 1181435, "postDate": "2021-02-01T22:04:32.033Z", "content": "Admin says: You must merge within the Kaggle platform before sharing any private competition information to be considered in adherence to competition rules.
\nSo they would be removed after this comp.
", "rawMarkdown": "Admin says: **You must merge within the Kaggle platform before sharing any private competition information to be considered in adherence to competition rules.**\n\nSo they would be removed after this comp.", "votes": 7, "replies": [ { "id": 1181514, "postDate": "2021-02-02T00:23:22.020Z", "content": "Where?
", "rawMarkdown": "Where? " }, { "id": 1181571, "postDate": "2021-02-02T01:34:58.210Z", "content": "here
\nhttps://www.kaggle.com/c/rfcx-species-audio-detection/discussion/197744
please note that working as a team before you merge does not necessarily include private sharing.
\nIf X, Y and Z agreed to team up later but continued to work on their solutions independently then, according to the rules, that seems to be fine.
\nI am not talking about this particular case but just saying it's possible, unlikely but possible.
", "rawMarkdown": "please note that working as a team before you merge does not necessarily include private sharing. \n\nIf X, Y and Z agreed to team up later but continued to work on their solutions independently then, according to the rules, that seems to be fine. \n\nI am not talking about this particular case but just saying it's possible, unlikely but possible.", "votes": 1 }, { "id": 1182491, "postDate": "2021-02-02T13:31:21.363Z", "content": "It's possible, but when the scores are too close, it's very likely private sharing.
\nNormally when solutions are very independent but still close in LB scores, merging would give significant boost on LB.
", "rawMarkdown": "It's possible, but when the scores are too close, it's very likely private sharing. \n\nNormally when solutions are very independent but still close in LB scores, merging would give significant boost on LB. ", "votes": 4 }, { "id": 1182570, "postDate": "2021-02-02T13:58:29.733Z", "content": "I am with you. No doubt.
\nHowever, IMHO, having a similar score, is NOT evidence of cheating? The same goes for cases where participants are from the same country, city, university, company, etc.
\nI believe that labeling someone as a cheater who may not be one is a very bad thing. I understand the frustration, I've been there too, but we must ensure that those who are coming to Kaggle with good intentions are never treated in a bad way. This is why I would rather refrain from an allegation of wrongdoing if I cannot guarantee that the wrongdoing actually happened.
", "rawMarkdown": "I am with you. No doubt. \n\nHowever, IMHO, having a similar score, is NOT evidence of cheating? The same goes for cases where participants are from the same country, city, university, company, etc. \n\nI believe that labeling someone as a cheater who may not be one is a very bad thing. I understand the frustration, I've been there too, but we must ensure that those who are coming to Kaggle with good intentions are never treated in a bad way. This is why I would rather refrain from an allegation of wrongdoing if I cannot guarantee that the wrongdoing actually happened. ", "votes": 1 }, { "id": 1182658, "postDate": "2021-02-02T14:30:51.713Z", "content": "\n\nWhile they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.
\n
Well, they were not labelled. I drew Kagglers' attention to an anomaly. They could just say \"Yeah, it is a coincidence. We would be happy to share our solutions with Kaggle if needed.\" Then nobody would argue.
", "rawMarkdown": "> While they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.\n\nWell, they were not labelled. I drew Kagglers' attention to an anomaly. They could just say \"Yeah, it is a coincidence. We would be happy to share our solutions with Kaggle if needed.\" Then nobody would argue." }, { "id": 1182715, "postDate": "2021-02-02T15:06:50.973Z", "content": "\n\nworking as a team before you merge does not necessarily include private sharing.
\n
How is working as a team defined in this case? A team that does not share anything is not a team.
\nI don't see where they would have shared publicly. I found nothing on Kaggle.
", "rawMarkdown": "> working as a team before you merge does not necessarily include private sharing. \n\nHow is working as a team defined in this case? A team that does not share anything is not a team.\n\nI don't see where they would have shared publicly. I found nothing on Kaggle.\n", "votes": 5 }, { "id": 1183326, "postDate": "2021-02-02T22:35:04.270Z", "content": "@cpmpml if several sides agree to merge but continue to work independently they are already a team in my eyes.
\nYou probably know this better than me that people ask some basic question before they merge (e.g. is your score based on SED or non-SED), to know whether they would benefit from a merge and ensemble. That is already shady but it is tolerated. Not? From that point, some decide to keep pushing independently for some time and they merge after they've tried their own ideas.
\nNote, I am not defending anyone here. I'm just trying to raise awareness that we are entering a risky zone here and that is why Kaggle wants us to inform them directly @inversion
", "rawMarkdown": "@cpmpml if several sides agree to merge but continue to work independently they are already a team in my eyes. \n\nYou probably know this better than me that people ask some basic question before they merge (e.g. is your score based on SED or non-SED), to know whether they would benefit from a merge and ensemble. That is already shady but it is tolerated. Not? From that point, some decide to keep pushing independently for some time and they merge after they've tried their own ideas. \n\nNote, I am not defending anyone here. I'm just trying to raise awareness that we are entering a risky zone here and that is why Kaggle wants us to inform them directly @inversion" }, { "id": 1183332, "postDate": "2021-02-02T22:42:09.170Z", "content": "I had similar post too 1.5 years ago during APTOS https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/105511
\nthey, as many others got kicked out but that didn't stop others from trying. I think Kaggle has to do a better job at making it visible that misconduct gets noticed and consequences apply
\nPerhaps we can have a pinned topic \"Cheaters get caught\" next to the \"external data\", \"looking for a team\" and \"new to ML\"
", "rawMarkdown": "I had similar post too 1.5 years ago during APTOS https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/105511\n\nthey, as many others got kicked out but that didn't stop others from trying. I think Kaggle has to do a better job at making it visible that misconduct gets noticed and consequences apply\n\nPerhaps we can have a pinned topic \"Cheaters get caught\" next to the \"external data\", \"looking for a team\" and \"new to ML\"" }, { "id": 1183765, "postDate": "2021-02-03T07:54:45.743Z", "content": "\n\nYou probably know this better than me that people ask some basic question before they merge (e.g. is your score based on SED or non-SED)
\n
That would be private sharing outside a team.
\nThat's why I do think that some private sharing is acceptable if the intent is to merge rapidly. But if you read the rules strictly then what you suggest is not allowed.
", "rawMarkdown": "> You probably know this better than me that people ask some basic question before they merge (e.g. is your score based on SED or non-SED)\n\nThat would be private sharing outside a team.\n\nThat's why I do think that some private sharing is acceptable if the intent is to merge rapidly. But if you read the rules strictly then what you suggest is not allowed.", "votes": 1 }, { "id": 1183810, "postDate": "2021-02-03T08:25:37.400Z", "content": "@cpmpm l fully agree. It seems we have very similar views. Just one of us sees the glass half full and the other half empty.
\nI agree with some private sharing is acceptable if the intent is to merge rapidly. We live with the fact that many such discussions do not end up with a merge and in that regard, I found it less problematic that some merges are postponed, given the amount of shared information is the same.
Interesting discussion. I appreciate it.
", "rawMarkdown": "@cpmpm l fully agree. It seems we have very similar views. Just one of us sees the glass half full and the other half empty.\n\nI agree with `some private sharing is acceptable if the intent is to merge rapidly`. We live with the fact that many such discussions do not end up with a merge and in that regard, I found it less problematic that some merges are postponed, given the amount of shared information is the same.\n\nInteresting discussion. I appreciate it.", "votes": 1 } ] }, { "id": 1181345, "postDate": "2021-02-01T19:53:41.827Z", "content": "Nice catch. As the other team, they decided to merge now that they were exposed ^^
", "rawMarkdown": "Nice catch. As the other team, they decided to merge now that they were exposed ^^", "votes": 5, "replies": [ { "id": 1181373, "postDate": "2021-02-01T20:37:37.423Z", "content": "They are 4 now. I expect the 5th person to join them.
", "rawMarkdown": "They are 4 now. I expect the 5th person to join them.", "votes": 2 }, { "id": 1181395, "postDate": "2021-02-01T21:07:24.387Z", "content": "Interesting team nameβ¦
", "rawMarkdown": "Interesting team name...", "votes": 6 }, { "id": 1181401, "postDate": "2021-02-01T21:16:02.280Z", "content": "I wonder why they merged. Each could get solo gold.
", "rawMarkdown": "I wonder why they merged. Each could get solo gold." }, { "id": 1181413, "postDate": "2021-02-01T21:25:32.630Z", "content": "If there is private sharing and the solution is close to money, there is a risk of not getting the money for some of the group members. So, I feel like merging is optimal as all members are more likely to win. Otherwise, if not merged, members outside the money cannot claim any right on the winner's prize legally.
", "rawMarkdown": "If there is private sharing and the solution is close to money, there is a risk of not getting the money for some of the group members. So, I feel like merging is optimal as all members are more likely to win. Otherwise, if not merged, members outside the money cannot claim any right on the winner's prize legally.", "votes": -2 }, { "id": 1181659, "postDate": "2021-02-02T03:29:26.787Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1181908, "postDate": "2021-02-02T07:44:18.263Z", "content": "@tolgadincer My comment was a sarcasm:)
", "rawMarkdown": "@tolgadincer My comment was a sarcasm:)", "votes": -1 }, { "id": 1182716, "postDate": "2021-02-02T15:08:02.070Z", "content": "Ahmet, That's why emoji were created: to convey info normally conveyed by body language. If you don't use them then no wonder some people miss the sarcasm ;)
", "rawMarkdown": "Ahmet, That's why emoji were created: to convey info normally conveyed by body language. If you don't use them then no wonder some people miss the sarcasm ;)", "votes": 2 } ] }, { "id": 1186447, "postDate": "2021-02-04T20:31:21.997Z", "content": "Here is our answer:
\nhttps://www.kaggle.com/c/rfcx-species-audio-detection/discussion/217000
Submitting at the same time is too suspicious to be a simple coincidence, I agree.
\nBut I think it would be better to write such information to admins for checking
", "rawMarkdown": "Submitting at the same time is too suspicious to be a simple coincidence, I agree.\n\nBut I think it would be better to write such information to admins for checking", "votes": 3, "replies": [ { "id": 1180488, "postDate": "2021-02-01T09:39:22.400Z", "content": "Kaggle admins tolerated private sharing in the past. Therefore I doubt sharing such information with them would be useful. They actually have more information than I have like login times, IPs etc.
", "rawMarkdown": "Kaggle admins tolerated private sharing in the past. Therefore I doubt sharing such information with them would be useful. They actually have more information than I have like login times, IPs etc.", "votes": 10 }, { "id": 1180737, "postDate": "2021-02-01T12:53:36.810Z", "content": "Itβs kind of disappointing to see that Kaggle, a data science platform, does not have a cheat detector.
", "rawMarkdown": "Itβs kind of disappointing to see that Kaggle, a data science platform, does not have a cheat detector.", "votes": 1 }, { "id": 1181255, "postDate": "2021-02-01T18:39:26.333Z", "content": "They do, and we have seen many (automatic?) LB removals in the past. But of course they are not telling how they detect suspicious activities.
", "rawMarkdown": "They do, and we have seen many (automatic?) LB removals in the past. But of course they are not telling how they detect suspicious activities. ", "votes": 1 }, { "id": 1182353, "postDate": "2021-02-02T12:26:58.703Z", "content": "Need more details; what information kaggle holds for user? like IP, Location, MAC etc?
", "rawMarkdown": "Need more details; what information kaggle holds for user? like IP, Location, MAC etc?" }, { "id": 1182687, "postDate": "2021-02-02T14:50:30.487Z", "content": "they did not, and they should not disclose it. Right now, a cheater does not know if he avoided the rules or not.
", "rawMarkdown": "they did not, and they should not disclose it. Right now, a cheater does not know if he avoided the rules or not." } ] }, { "id": 1182645, "postDate": "2021-02-02T14:26:18.863Z", "content": "If you didn't catch them, I guess they won't merge and get all banned at the end of this competition π€
\nNow they merged, I'm wondering if they will be banned.
They wouldn't be banned as long as they don't submit from same IP. I have never seen an example of private sharing penalized on Kaggle.
", "rawMarkdown": "They wouldn't be banned as long as they don't submit from same IP. I have never seen an example of private sharing penalized on Kaggle.", "votes": -1 }, { "id": 1182672, "postDate": "2021-02-02T14:38:10.807Z", "content": "Thats good news. Kagglers! All teams could private sharing and only need to merge at merger deadline.
", "rawMarkdown": "Thats good news. Kagglers! All teams could private sharing and only need to merge at merger deadline. ", "votes": -2 }, { "id": 1182673, "postDate": "2021-02-02T14:39:44.563Z", "content": "@aerdem4 Yes, people are removed from competitions because of private sharing.
", "rawMarkdown": "@aerdem4 Yes, people are removed from competitions because of private sharing. ", "votes": 3 }, { "id": 1182679, "postDate": "2021-02-02T14:44:47.673Z", "content": "Well, at least not in:
\nhttps://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard
\nhttps://www.kaggle.com/c/imet-2019-fgvc6/leaderboard
Do you know why a top-5 solution got that many downvotes?
\nhttps://www.kaggle.com/c/imet-2019-fgvc6/discussion/95223
Sorry my mistake, private sharing no need to merge. multi solo golds or prizes is best for them.
", "rawMarkdown": "Sorry my mistake, private sharing no need to merge. multi solo golds or prizes is best for them. ", "votes": -1 }, { "id": 1183724, "postDate": "2021-02-03T07:16:38.420Z", "content": "@aerdem4
\nYes, I finish https://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard at 11st position that time. There are many last time rapid boosted submission by Russian and none removed.
\nAnd one year later, I finish RSNA STR at 12st postion and there are many last time rapid boosted submission by Chinese and all of them are removed.
I am thinking.. are kaggle's anti cheat system advanced over time or the difference between Chinese or Russian, or evidence are not enough that time.
", "rawMarkdown": "@aerdem4 \nYes, I finish `https://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard` at 11st position that time. There are many last time rapid boosted submission by Russian and none removed.\nAnd one year later, I finish RSNA STR at 12st postion and there are many last time rapid boosted submission by Chinese and all of them are removed.\n\nI am thinking.. are kaggle's anti cheat system advanced over time or the difference between Chinese or Russian, or evidence are not enough that time.", "votes": 2 }, { "id": 1183726, "postDate": "2021-02-03T07:17:15.580Z", "content": "Thanks for you still remember this.
", "rawMarkdown": "Thanks for you still remember this.", "votes": 1 }, { "id": 1183744, "postDate": "2021-02-03T07:36:02.790Z", "content": "@steamedsheep how experienced were those removed from RSNA? IMO, a more experienced kaggler who decides to cheat is less likely to be caught as they will probably be aware of many cases where people got caught.
\nJust adding a little bit of noise to your submission, e.g. adding another batch to the training of your models gives different outputs with likely the same results. Not?
", "rawMarkdown": "@steamedsheep how experienced were those removed from RSNA? IMO, a more experienced kaggler who decides to cheat is less likely to be caught as they will probably be aware of many cases where people got caught. \n\nJust adding a little bit of noise to your submission, e.g. adding another batch to the training of your models gives different outputs with likely the same results. Not?" } ] }, { "id": 1190100, "postDate": "2021-02-07T13:38:32.753Z", "content": "Same mother?
", "rawMarkdown": "Same mother? " }, { "id": 1190102, "postDate": "2021-02-07T13:39:55.510Z", "content": "I think Kaggle do its job great in recent years.
\nIf they get gold, let them share solution so that we can have a look. Otherwise, it can be just a coincidence (or not).
Paranormal activities π
", "rawMarkdown": "Paranormal activities π" }, { "id": 1181837, "postDate": "2021-02-02T06:32:23.580Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1180438, "author_name": "Dieter", "author_url": "", "post_date": "2021-02-01T08:56:42.837000", "content": "same university btw π
", "votes": 12, "replies": [] }, { "id": 1181435, "author_name": "ONODERA", "author_url": "", "post_date": "2021-02-01T22:04:32.033000", "content": "Admin says: You must merge within the Kaggle platform before sharing any private competition information to be considered in adherence to competition rules.
\nSo they would be removed after this comp.
", "votes": 7, "replies": [ { "id": 1181514, "author_name": "CPMP", "author_url": "", "post_date": "2021-02-02T00:23:22.020000", "content": "Where?
", "votes": 0, "replies": [] }, { "id": 1181571, "author_name": "ONODERA", "author_url": "", "post_date": "2021-02-02T01:34:58.210000", "content": "here
\nhttps://www.kaggle.com/c/rfcx-species-audio-detection/discussion/197744
please note that working as a team before you merge does not necessarily include private sharing.
\nIf X, Y and Z agreed to team up later but continued to work on their solutions independently then, according to the rules, that seems to be fine.
\nI am not talking about this particular case but just saying it's possible, unlikely but possible.
", "votes": 1, "replies": [] }, { "id": 1182491, "author_name": "Serigne ", "author_url": "", "post_date": "2021-02-02T13:31:21.363000", "content": "It's possible, but when the scores are too close, it's very likely private sharing.
\nNormally when solutions are very independent but still close in LB scores, merging would give significant boost on LB.
", "votes": 4, "replies": [] }, { "id": 1182570, "author_name": "Miroslav Valan", "author_url": "", "post_date": "2021-02-02T13:58:29.733000", "content": "I am with you. No doubt.
\nHowever, IMHO, having a similar score, is NOT evidence of cheating? The same goes for cases where participants are from the same country, city, university, company, etc.
\nI believe that labeling someone as a cheater who may not be one is a very bad thing. I understand the frustration, I've been there too, but we must ensure that those who are coming to Kaggle with good intentions are never treated in a bad way. This is why I would rather refrain from an allegation of wrongdoing if I cannot guarantee that the wrongdoing actually happened.
", "votes": 1, "replies": [] }, { "id": 1182658, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2021-02-02T14:30:51.713000", "content": "\n\nWhile they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.
\n
Well, they were not labelled. I drew Kagglers' attention to an anomaly. They could just say \"Yeah, it is a coincidence. We would be happy to share our solutions with Kaggle if needed.\" Then nobody would argue.
", "votes": 0, "replies": [] }, { "id": 1182715, "author_name": "CPMP", "author_url": "", "post_date": "2021-02-02T15:06:50.973000", "content": "\n\nworking as a team before you merge does not necessarily include private sharing.
\n
How is working as a team defined in this case? A team that does not share anything is not a team.
\nI don't see where they would have shared publicly. I found nothing on Kaggle.
", "votes": 5, "replies": [] }, { "id": 1183326, "author_name": "Miroslav Valan", "author_url": "", "post_date": "2021-02-02T22:35:04.270000", "content": "@cpmpml if several sides agree to merge but continue to work independently they are already a team in my eyes.
\nYou probably know this better than me that people ask some basic question before they merge (e.g. is your score based on SED or non-SED), to know whether they would benefit from a merge and ensemble. That is already shady but it is tolerated. Not? From that point, some decide to keep pushing independently for some time and they merge after they've tried their own ideas.
\nNote, I am not defending anyone here. I'm just trying to raise awareness that we are entering a risky zone here and that is why Kaggle wants us to inform them directly @inversion
", "votes": 0, "replies": [] }, { "id": 1183332, "author_name": "Miroslav Valan", "author_url": "", "post_date": "2021-02-02T22:42:09.170000", "content": "I had similar post too 1.5 years ago during APTOS https://www.kaggle.com/c/aptos2019-blindness-detection/discussion/105511
\nthey, as many others got kicked out but that didn't stop others from trying. I think Kaggle has to do a better job at making it visible that misconduct gets noticed and consequences apply
\nPerhaps we can have a pinned topic \"Cheaters get caught\" next to the \"external data\", \"looking for a team\" and \"new to ML\"
", "votes": 0, "replies": [] }, { "id": 1183765, "author_name": "CPMP", "author_url": "", "post_date": "2021-02-03T07:54:45.743000", "content": "\n\nYou probably know this better than me that people ask some basic question before they merge (e.g. is your score based on SED or non-SED)
\n
That would be private sharing outside a team.
\nThat's why I do think that some private sharing is acceptable if the intent is to merge rapidly. But if you read the rules strictly then what you suggest is not allowed.
", "votes": 1, "replies": [] }, { "id": 1183810, "author_name": "Miroslav Valan", "author_url": "", "post_date": "2021-02-03T08:25:37.400000", "content": "@cpmpm l fully agree. It seems we have very similar views. Just one of us sees the glass half full and the other half empty.
\nI agree with some private sharing is acceptable if the intent is to merge rapidly. We live with the fact that many such discussions do not end up with a merge and in that regard, I found it less problematic that some merges are postponed, given the amount of shared information is the same.
Interesting discussion. I appreciate it.
", "votes": 1, "replies": [] } ] }, { "id": 1181345, "author_name": "Theo Viel", "author_url": "", "post_date": "2021-02-01T19:53:41.827000", "content": "Nice catch. As the other team, they decided to merge now that they were exposed ^^
", "votes": 5, "replies": [ { "id": 1181373, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2021-02-01T20:37:37.423000", "content": "They are 4 now. I expect the 5th person to join them.
", "votes": 2, "replies": [] }, { "id": 1181395, "author_name": "Psi", "author_url": "", "post_date": "2021-02-01T21:07:24.387000", "content": "Interesting team nameβ¦
", "votes": 6, "replies": [] }, { "id": 1181401, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2021-02-01T21:16:02.280000", "content": "I wonder why they merged. Each could get solo gold.
", "votes": 0, "replies": [] }, { "id": 1181413, "author_name": "Tolga", "author_url": "", "post_date": "2021-02-01T21:25:32.630000", "content": "If there is private sharing and the solution is close to money, there is a risk of not getting the money for some of the group members. So, I feel like merging is optimal as all members are more likely to win. Otherwise, if not merged, members outside the money cannot claim any right on the winner's prize legally.
", "votes": -2, "replies": [] }, { "id": 1181659, "author_name": "", "author_url": "", "post_date": "2021-02-02T03:29:26.787000", "content": "", "votes": 0, "replies": [] }, { "id": 1181908, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2021-02-02T07:44:18.263000", "content": "@tolgadincer My comment was a sarcasm:)
", "votes": -1, "replies": [] }, { "id": 1182716, "author_name": "CPMP", "author_url": "", "post_date": "2021-02-02T15:08:02.070000", "content": "Ahmet, That's why emoji were created: to convey info normally conveyed by body language. If you don't use them then no wonder some people miss the sarcasm ;)
", "votes": 2, "replies": [] } ] }, { "id": 1186447, "author_name": "DimaSakovich", "author_url": "", "post_date": "2021-02-04T20:31:21.997000", "content": "Here is our answer:
\nhttps://www.kaggle.com/c/rfcx-species-audio-detection/discussion/217000
Submitting at the same time is too suspicious to be a simple coincidence, I agree.
\nBut I think it would be better to write such information to admins for checking
", "votes": 3, "replies": [ { "id": 1180488, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2021-02-01T09:39:22.400000", "content": "Kaggle admins tolerated private sharing in the past. Therefore I doubt sharing such information with them would be useful. They actually have more information than I have like login times, IPs etc.
", "votes": 10, "replies": [] }, { "id": 1180737, "author_name": "Tolga", "author_url": "", "post_date": "2021-02-01T12:53:36.810000", "content": "Itβs kind of disappointing to see that Kaggle, a data science platform, does not have a cheat detector.
", "votes": 1, "replies": [] }, { "id": 1181255, "author_name": "Pascal Pfeiffer", "author_url": "", "post_date": "2021-02-01T18:39:26.333000", "content": "They do, and we have seen many (automatic?) LB removals in the past. But of course they are not telling how they detect suspicious activities.
", "votes": 1, "replies": [] }, { "id": 1182353, "author_name": "Adnan Zaidi", "author_url": "", "post_date": "2021-02-02T12:26:58.703000", "content": "Need more details; what information kaggle holds for user? like IP, Location, MAC etc?
", "votes": 0, "replies": [] }, { "id": 1182687, "author_name": "Miroslav Valan", "author_url": "", "post_date": "2021-02-02T14:50:30.487000", "content": "they did not, and they should not disclose it. Right now, a cheater does not know if he avoided the rules or not.
", "votes": 0, "replies": [] } ] }, { "id": 1182645, "author_name": "yuanzhe zhou", "author_url": "", "post_date": "2021-02-02T14:26:18.863000", "content": "If you didn't catch them, I guess they won't merge and get all banned at the end of this competition π€
\nNow they merged, I'm wondering if they will be banned.
They wouldn't be banned as long as they don't submit from same IP. I have never seen an example of private sharing penalized on Kaggle.
", "votes": -1, "replies": [] }, { "id": 1182672, "author_name": "MaChaogong", "author_url": "", "post_date": "2021-02-02T14:38:10.807000", "content": "Thats good news. Kagglers! All teams could private sharing and only need to merge at merger deadline.
", "votes": -2, "replies": [] }, { "id": 1182673, "author_name": "inversion", "author_url": "", "post_date": "2021-02-02T14:39:44.563000", "content": "@aerdem4 Yes, people are removed from competitions because of private sharing.
", "votes": 3, "replies": [] }, { "id": 1182679, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2021-02-02T14:44:47.673000", "content": "Well, at least not in:
\nhttps://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard
\nhttps://www.kaggle.com/c/imet-2019-fgvc6/leaderboard
Do you know why a top-5 solution got that many downvotes?
\nhttps://www.kaggle.com/c/imet-2019-fgvc6/discussion/95223
Sorry my mistake, private sharing no need to merge. multi solo golds or prizes is best for them.
", "votes": -1, "replies": [] }, { "id": 1183724, "author_name": "sheep", "author_url": "", "post_date": "2021-02-03T07:16:38.420000", "content": "@aerdem4
\nYes, I finish https://www.kaggle.com/c/imaterialist-fashion-2019-FGVC6/leaderboard at 11st position that time. There are many last time rapid boosted submission by Russian and none removed.
\nAnd one year later, I finish RSNA STR at 12st postion and there are many last time rapid boosted submission by Chinese and all of them are removed.
I am thinking.. are kaggle's anti cheat system advanced over time or the difference between Chinese or Russian, or evidence are not enough that time.
", "votes": 2, "replies": [] }, { "id": 1183726, "author_name": "sheep", "author_url": "", "post_date": "2021-02-03T07:17:15.580000", "content": "Thanks for you still remember this.
", "votes": 1, "replies": [] }, { "id": 1183744, "author_name": "Miroslav Valan", "author_url": "", "post_date": "2021-02-03T07:36:02.790000", "content": "@steamedsheep how experienced were those removed from RSNA? IMO, a more experienced kaggler who decides to cheat is less likely to be caught as they will probably be aware of many cases where people got caught.
\nJust adding a little bit of noise to your submission, e.g. adding another batch to the training of your models gives different outputs with likely the same results. Not?
", "votes": 0, "replies": [] } ] }, { "id": 1190100, "author_name": "Mukharbek Organokov", "author_url": "", "post_date": "2021-02-07T13:38:32.753000", "content": "Same mother?
", "votes": 0, "replies": [] }, { "id": 1190102, "author_name": "Mukharbek Organokov", "author_url": "", "post_date": "2021-02-07T13:39:55.510000", "content": "I think Kaggle do its job great in recent years.
\nIf they get gold, let them share solution so that we can have a look. Otherwise, it can be just a coincidence (or not).
Paranormal activities π
", "votes": 0, "replies": [] }, { "id": 1181837, "author_name": "", "author_url": "", "post_date": "2021-02-02T06:32:23.580000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1180301": "\n\n**#11** and **#14** are at similar score and **submit exactly at the same time**. They are from the same country and follow each other. Apologies if it is a coincidence but submitting at the same time is another unlikely thing.\n\nWhile they look suspicious, they can also be 100% legit. I think Kaggle should have such rule based filters and randomly check people to explain their solution on ongoing competitions.", "1180438": "same university btw π", "1181435": "Admin says: **You must merge within the Kaggle platform before sharing any private competition information to be considered in adherence to competition rules.**\n\nSo they would be removed after this comp.", "1181345": "Nice catch. As the other team, they decided to merge now that they were exposed ^^", "1186447": "Here is our answer:\nhttps://www.kaggle.com/c/rfcx-species-audio-detection/discussion/217000", "1180486": "Submitting at the same time is too suspicious to be a simple coincidence, I agree.\n\nBut I think it would be better to write such information to admins for checking", "1182645": "If you didn't catch them, I guess they won't merge and get all banned at the end of this competition π€\nNow they merged, I'm wondering if they will be banned.", "1190100": "Same mother? ", "1190102": "I think Kaggle do its job great in recent years.\nIf they get gold, let them share solution so that we can have a look. Otherwise, it can be just a coincidence (or not). ", "1181251": "Paranormal activities π", "1181837": "" } }