![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F5168115%2F86706e15eb9cd6303ff64fb1a4d76aa2%2FScreenshot_2020-10-25%20hypjudy%20Kaggle.png?generation=1603687764118520&alt=media\")
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F5168115%2F20d57447cadd7b7a364f7d2e43c25835%2FScreenshot_2020-10-25%20wjbkimberly%20Kaggle.png?generation=1603687647740089&alt=media\")
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F5168115%2F9f2f9c764f8436de4e1ed7fc88258b31%2Fimage(6).png?generation=1603687724547098&alt=media\")
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F5168115%2F6c501af669db13de76fd2e43c7b1457c%2Fsct.png?generation=1603687863906568&alt=media\")
We worked extremely hard in this competition and yet still can’t break LB 0.16.
\nIt’s just mesmerizing to witness some talented competitors who climbed the leaderboard within a few submissions during their first competition.
\nJust can’t wait to read their excellent solutions ;)
", "messages": [ { "id": 1060343, "postDate": "2020-10-26T04:44:54.987Z", "content": "We worked extremely hard in this competition and yet still can’t break LB 0.16.
\nIt’s just mesmerizing to witness some talented competitors who climbed the leaderboard within a few submissions during their first competition.
\nJust can’t wait to read their excellent solutions ;)
", "rawMarkdown": "We worked extremely hard in this competition and yet still can’t break LB 0.16.\n\nIt’s just mesmerizing to witness some talented competitors who climbed the leaderboard within a few submissions during their first competition.\n\nJust can’t wait to read their excellent solutions ;)", "votes": 46 }, { "id": 1060344, "postDate": "2020-10-26T04:51:19.270Z", "content": "
\n
\n
\n
Achieving 0.15 after just 7 submissions looks suspicious…
", "rawMarkdown": "Achieving 0.15 after just 7 submissions looks suspicious... ", "votes": 3 }, { "id": 1060410, "postDate": "2020-10-26T07:04:05.723Z", "content": "notice that its username is 'wjbmatrix68' which has the same prefix 'wjb' as the 'wjbkimberly' in the first team.
", "rawMarkdown": "notice that its username is 'wjbmatrix68' which has the same prefix 'wjb' as the 'wjbkimberly' in the first team.", "votes": 7 }, { "id": 1060423, "postDate": "2020-10-26T07:14:36.863Z", "content": "@philculliton please have a look at the above - hopefully, Kaggle anti-fraud mechanism can check this thoroughly
", "rawMarkdown": "@philculliton please have a look at the above - hopefully, Kaggle anti-fraud mechanism can check this thoroughly", "votes": 5 }, { "id": 1060432, "postDate": "2020-10-26T07:30:03.503Z", "content": "\"Achieving 0.15 after just 7 submissions looks suspicious…\" actually just 3 submissions
", "rawMarkdown": "\"Achieving 0.15 after just 7 submissions looks suspicious…\" actually just 3 submissions", "votes": 5 }, { "id": 1060716, "postDate": "2020-10-26T13:53:08.627Z", "content": "@underwearfitting if they were committing fraud or if they were sharing code against the rules, why would they leave any evidence against them? Why would they leave a prefix linking the two accounts?
", "rawMarkdown": "@underwearfitting ***if*** they were committing fraud or if they were sharing code against the rules, why would they leave any evidence against them? Why would they leave a prefix linking the two accounts?", "votes": 1 }, { "id": 1061253, "postDate": "2020-10-26T21:26:15.013Z", "content": "@darkcube In fact, I never said they are committing fraud. I just posted some screenshots. It is left for you to interpret.
", "rawMarkdown": "@darkcube In fact, I never said they are committing fraud. I just posted some screenshots. It is left for you to interpret. ", "votes": 1 }, { "id": 1061261, "postDate": "2020-10-26T21:45:26.657Z", "content": "What I meant is why would they do something that would draw suspicion.
\nIt literally makes no sense.
", "rawMarkdown": "What I meant is why would they do something that would draw suspicion.\n\nIt literally makes no sense." } ] }, { "id": 1060599, "postDate": "2020-10-26T11:39:03.490Z", "content": "Submission count doesn't matter in this competition. CV/LB is very correlated. You basically don't need to submit after noticing that.
", "rawMarkdown": "Submission count doesn't matter in this competition. CV/LB is very correlated. You basically don't need to submit after noticing that.", "votes": 6, "replies": [ { "id": 1060625, "postDate": "2020-10-26T12:09:46.223Z", "content": "So you guys got CV score around 0.190? That's a good CV score to us.
\nActually our CV score is only around 0.20 😮
I don't know why you're getting downvoted.
\nMy first submission that didn't time out was 0.159 (CV 0.195). CV score for my current LB best (0.150) is 0.183. For me, CV is off by LB by about 0.03- but performance improvements are correlated.
", "rawMarkdown": "I don't know why you're getting downvoted. \n\nMy first submission that didn't time out was 0.159 (CV 0.195). CV score for my current LB best (0.150) is 0.183. For me, CV is off by LB by about 0.03- but performance improvements are correlated. ", "votes": 9 }, { "id": 1060634, "postDate": "2020-10-26T12:24:12.553Z", "content": "\n\nI don't know why you're getting downvoted.
\n
Don't try understanding downvotes, it is a bottomless sink.
", "rawMarkdown": "> I don't know why you're getting downvoted. \n\nDon't try understanding downvotes, it is a bottomless sink.", "votes": 7 }, { "id": 1060636, "postDate": "2020-10-26T12:26:06.530Z", "content": "@haqishen I didn't say \"the same\", I said \"correlated\".
\n@vaillant thanks for the example. I hope downvoters can explain why making a lot of submissions in this competition is useful.
", "rawMarkdown": "@haqishen I didn't say \"the same\", I said \"correlated\".\n\n@vaillant thanks for the example. I hope downvoters can explain why making a lot of submissions in this competition is useful.", "votes": 2 }, { "id": 1060638, "postDate": "2020-10-26T12:26:56.073Z", "content": "@haqishen Correlated does not mean they are equal.
", "rawMarkdown": "@haqishen Correlated does not mean they are equal. ", "votes": 1 }, { "id": 1060639, "postDate": "2020-10-26T12:29:33.493Z", "content": "i myself got 0.154 after 3 subs from the very early of this competition so it's understandable. anything can happen and i trust the kaggle team. just wait till tomorrow and we will know if those guys are legit or not
", "rawMarkdown": "i myself got 0.154 after 3 subs from the very early of this competition so it's understandable. anything can happen and i trust the kaggle team. just wait till tomorrow and we will know if those guys are legit or not", "votes": 10 }, { "id": 1060988, "postDate": "2020-10-26T17:16:01.920Z", "content": "I had a quite big CV / LB gap at earlier (0.02), turns out there was a mistake in my implementation, which I adapted from here. Now I have a gap of 0.002~
\nI was able to fix the issue by making sure submitting 0.5 everywhere gives ln(2) = 0.69314, the normalization of the image weights was apparently wrong, if that helps anyone.
2/3 teams didn't choose their best submission, so they don't need to publish solution 😂😂😂
", "rawMarkdown": "2/3 teams didn't choose their best submission, so they don't need to publish solution 😂😂😂", "votes": 3, "replies": [ { "id": 1061405, "postDate": "2020-10-27T01:53:43.520Z", "content": "Sadly we won't be able to read their excellent solutions ;(
\nThey've deleted the accounts.
There are no real deletion in database, I am curious where their submission are comes from, maybe admin can tell.
", "rawMarkdown": "There are no real deletion in database, I am curious where their submission are comes from, maybe admin can tell. ", "votes": 2 } ] }, { "id": 1061423, "postDate": "2020-10-27T02:21:39.430Z", "content": "Now these 'talent' delete their account in order to avoid private share investigation.
\nI am highly recommand @philculliton to do similiary test between all submission. It's time to figure out where private share are comes from and tell us the source of their 'talent'.
\nI guess if this discussion not exist, they may finish their compeition 'normally'.
https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/193338
\nThey have since deleted their GitHub account, with over 500 stars too.
", "rawMarkdown": "https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/193338\n\nThey have since deleted their GitHub account, with over 500 stars too. ", "votes": 2 }, { "id": 1061519, "postDate": "2020-10-27T03:59:29.420Z", "content": "Also they changed their names :D. I am pretty sure @steamedsheep will get a solo gold if they were punished. Congrats for you solo gold finish!
", "rawMarkdown": "Also they changed their names :D. I am pretty sure @steamedsheep will get a solo gold if they were punished. Congrats for you solo gold finish!", "votes": 4 }, { "id": 1062014, "postDate": "2020-10-27T14:00:16.960Z", "content": "Suspicious to say the least. 2 top-10 solo competitors with potential ties to the 5-man 2nd place team who deleted their accounts at the conclusion of the competition.
\nIt seems like they were being used to submit dummy submissions perhaps at an attempt to get around the daily submission limit. Of course, this is all just speculation.
", "rawMarkdown": "Suspicious to say the least. 2 top-10 solo competitors with potential ties to the 5-man 2nd place team who deleted their accounts at the conclusion of the competition. \n\nIt seems like they were being used to submit dummy submissions perhaps at an attempt to get around the daily submission limit. Of course, this is all just speculation. ", "votes": 4 }, { "id": 1062024, "postDate": "2020-10-27T14:08:20.487Z", "content": "No. 13 team on public LB also deleted the account, in addition to No. 5 and No. 8 (cc @juliaelliott )
\nI also highly suspect these accounts were used by MSM team (public No. 1 team) to circumvent the daily submission limit. I noticed MSM was using \"fast submission\" throughout the competition. My guess is that they needed more subs in the end to debug \"real submission\" (even though CV and LB are highly correlated in this competition)
", "rawMarkdown": "No. 13 team on public LB also deleted the account, in addition to No. 5 and No. 8 (cc @juliaelliott )\n\nI also highly suspect these accounts were used by MSM team (public No. 1 team) to circumvent the daily submission limit. I noticed MSM was using \"fast submission\" throughout the competition. My guess is that they needed more subs in the end to debug \"real submission\" (even though CV and LB are highly correlated in this competition)", "votes": 4 } ] }, { "id": 1060357, "postDate": "2020-10-26T05:29:08.760Z", "content": "I love and hate kaggle.
\nThe more I learn, the greater frustration I have, mostly by realizing the huge distance between the average (eg me) and the smart 😔😔
Achieving 0.15 after just 7 submissions.Five of them were submitted by the day before.It doesn't seem that he has worked hard. Otherwise, he will try to submit before, unless he is sure that he will get good results, so there is no need to submit before.Of course, it's just my guess…
", "rawMarkdown": "Achieving 0.15 after just 7 submissions.Five of them were submitted by the day before.It doesn't seem that he has worked hard. Otherwise, he will try to submit before, unless he is sure that he will get good results, so there is no need to submit before.Of course, it's just my guess...", "votes": 1, "replies": [ { "id": 1060544, "postDate": "2020-10-26T09:41:52.030Z", "content": "Actually just 3 submissions, not 7.
", "rawMarkdown": "Actually just 3 submissions, not 7.", "votes": 2 } ] }, { "id": 1060661, "postDate": "2020-10-26T12:55:14.123Z", "content": "that Chinese guy in 1st place. He was alone for a while and then all of a sudden he had 4 more teammates. just about when the merging deadline was coming to a close. That was a bit suspicious.
\nEdit: why tf am I getting downvoted?
", "rawMarkdown": "that Chinese guy in 1st place. He was alone for a while and then all of a sudden he had 4 more teammates. just about when the merging deadline was coming to a close. That was a bit suspicious.\n\nEdit: why tf am I getting downvoted?", "votes": -12, "replies": [ { "id": 1060686, "postDate": "2020-10-26T13:18:33.290Z", "content": "I don't know WHAT guy you are. But I do know why you're getting downvoted.
", "rawMarkdown": "I don't know WHAT guy you are. But I do know why you're getting downvoted.", "votes": 5 }, { "id": 1060711, "postDate": "2020-10-26T13:48:58.290Z", "content": "are you willing to tell me?
", "rawMarkdown": "are you willing to tell me?", "votes": -3 }, { "id": 1060732, "postDate": "2020-10-26T14:11:08.313Z", "content": "Hey mate, noticed you getting downvoted quite a ton on this forum in general. It is generally better not to make assumptions and say things like \"TTA doesn't sound like it improves your score much\" or saying that someone is suspicious without evidence - newbie teams win competitions all the time, but true investigative journalism exists too, and Gilles' forum post is a great example. https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175156
", "rawMarkdown": "Hey mate, noticed you getting downvoted quite a ton on this forum in general. It is generally better not to make assumptions and say things like \"TTA doesn't sound like it improves your score much\" or saying that someone is suspicious without evidence - newbie teams win competitions all the time, but true investigative journalism exists too, and Gilles' forum post is a great example. https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175156", "votes": 2 }, { "id": 1060762, "postDate": "2020-10-26T14:32:37.677Z", "content": "\n\nIt is generally better not to make assumptions and say things like saying that someone is suspicious without evidence
\n
if you have evidence then he's not a suspect anymore. He's been accused. I wasn't accusing anyone, I was just raising questions. Why would he choose to merge with just a week to go? Why are they all novices? And if they know each other why did they merge so late?
\n\n\nIt is generally better not to make assumptions and say things like \"TTA doesn't sound like it improves your score much\"
\n
Bro, it is generally accepted that hypothesizing is very much okay.
", "rawMarkdown": "> It is generally better not to make assumptions and say things like saying that someone is suspicious without evidence\n\nif you have evidence then he's not a suspect anymore. He's been accused. I wasn't accusing anyone, I was just raising questions. Why would he choose to merge with just a week to go? Why are they all novices? And if they know each other why did they merge so late?\n\n> It is generally better not to make assumptions and say things like \"TTA doesn't sound like it improves your score much\"\n\nBro, it is generally accepted that hypothesizing is very much okay.", "votes": -3 }, { "id": 1060877, "postDate": "2020-10-26T15:39:24.023Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1060906, "postDate": "2020-10-26T15:58:21.587Z", "content": "I know, I'm just giving you my side of the story.
", "rawMarkdown": "I know, I'm just giving you my side of the story.", "votes": -3 }, { "id": 1061251, "postDate": "2020-10-26T21:23:49.447Z", "content": "How do you know he's Chinese??? If I put a gaggle as my profile picture, am I a gaggle???😑😑
", "rawMarkdown": "How do you know he's Chinese??? If I put a gaggle as my profile picture, am I a gaggle???😑😑", "votes": 2 }, { "id": 1061252, "postDate": "2020-10-26T21:25:11.807Z", "content": "Haha @underwearfitting hilarious comment, never heard them being called \"gaggle\" but it's a perfect name.
\nIt also does say in his profile, \"Beijing China\"
@stanleyjzheng Kaggle's default pic is gaggle for a reason haha. Same logic here, if I put Tokyo on my profile, does that make me Japanese?😂
", "rawMarkdown": "@stanleyjzheng Kaggle's default pic is gaggle for a reason haha. Same logic here, if I put Tokyo on my profile, does that make me Japanese?😂", "votes": 1 }, { "id": 1061263, "postDate": "2020-10-26T21:49:03.607Z", "content": "I mean I just said he was Chinese as a reference. There is no need to pull an investigation about it. Just as I'd say the one with the starry night a profile or the one with a baby.
", "rawMarkdown": "I mean I just said he was Chinese as a reference. There is no need to pull an investigation about it. Just as I'd say the one with the starry night a profile or the one with a baby." } ] }, { "id": 1061694, "postDate": "2020-10-27T08:08:37.113Z", "content": "Hey @haqishen, can you please provide the link to the external data you have used. I am searching for additional PE dataset. [\"All Data Are Ext\"]
Sure, here you go
\nhttps://www.kaggle.com/c/deepfake-detection-challenge/discussion/157983
\n
\n
\n
Achieving 0.15 after just 7 submissions looks suspicious…
", "votes": 3, "replies": [] }, { "id": 1060410, "author_name": "sin", "author_url": "", "post_date": "2020-10-26T07:04:05.723000", "content": "notice that its username is 'wjbmatrix68' which has the same prefix 'wjb' as the 'wjbkimberly' in the first team.
", "votes": 7, "replies": [] }, { "id": 1060423, "author_name": "narsil (jobs-in-data.com)", "author_url": "", "post_date": "2020-10-26T07:14:36.863000", "content": "@philculliton please have a look at the above - hopefully, Kaggle anti-fraud mechanism can check this thoroughly
", "votes": 5, "replies": [] }, { "id": 1060432, "author_name": "DungNB", "author_url": "", "post_date": "2020-10-26T07:30:03.503000", "content": "\"Achieving 0.15 after just 7 submissions looks suspicious…\" actually just 3 submissions
", "votes": 5, "replies": [] }, { "id": 1060716, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T13:53:08.627000", "content": "@underwearfitting if they were committing fraud or if they were sharing code against the rules, why would they leave any evidence against them? Why would they leave a prefix linking the two accounts?
", "votes": 1, "replies": [] }, { "id": 1061253, "author_name": "sin", "author_url": "", "post_date": "2020-10-26T21:26:15.013000", "content": "@darkcube In fact, I never said they are committing fraud. I just posted some screenshots. It is left for you to interpret.
", "votes": 1, "replies": [] }, { "id": 1061261, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T21:45:26.657000", "content": "What I meant is why would they do something that would draw suspicion.
\nIt literally makes no sense.
", "votes": 0, "replies": [] } ] }, { "id": 1060599, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2020-10-26T11:39:03.490000", "content": "Submission count doesn't matter in this competition. CV/LB is very correlated. You basically don't need to submit after noticing that.
", "votes": 6, "replies": [ { "id": 1060625, "author_name": "Qishen Ha", "author_url": "", "post_date": "2020-10-26T12:09:46.223000", "content": "So you guys got CV score around 0.190? That's a good CV score to us.
\nActually our CV score is only around 0.20 😮
I don't know why you're getting downvoted.
\nMy first submission that didn't time out was 0.159 (CV 0.195). CV score for my current LB best (0.150) is 0.183. For me, CV is off by LB by about 0.03- but performance improvements are correlated.
", "votes": 9, "replies": [] }, { "id": 1060634, "author_name": "CPMP", "author_url": "", "post_date": "2020-10-26T12:24:12.553000", "content": "\n\nI don't know why you're getting downvoted.
\n
Don't try understanding downvotes, it is a bottomless sink.
", "votes": 7, "replies": [] }, { "id": 1060636, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2020-10-26T12:26:06.530000", "content": "@haqishen I didn't say \"the same\", I said \"correlated\".
\n@vaillant thanks for the example. I hope downvoters can explain why making a lot of submissions in this competition is useful.
", "votes": 2, "replies": [] }, { "id": 1060638, "author_name": "CPMP", "author_url": "", "post_date": "2020-10-26T12:26:56.073000", "content": "@haqishen Correlated does not mean they are equal.
", "votes": 1, "replies": [] }, { "id": 1060639, "author_name": "DatNT", "author_url": "", "post_date": "2020-10-26T12:29:33.493000", "content": "i myself got 0.154 after 3 subs from the very early of this competition so it's understandable. anything can happen and i trust the kaggle team. just wait till tomorrow and we will know if those guys are legit or not
", "votes": 10, "replies": [] }, { "id": 1060988, "author_name": "Theo Viel", "author_url": "", "post_date": "2020-10-26T17:16:01.920000", "content": "I had a quite big CV / LB gap at earlier (0.02), turns out there was a mistake in my implementation, which I adapted from here. Now I have a gap of 0.002~
\nI was able to fix the issue by making sure submitting 0.5 everywhere gives ln(2) = 0.69314, the normalization of the image weights was apparently wrong, if that helps anyone.
2/3 teams didn't choose their best submission, so they don't need to publish solution 😂😂😂
", "votes": 3, "replies": [ { "id": 1061405, "author_name": "Qishen Ha", "author_url": "", "post_date": "2020-10-27T01:53:43.520000", "content": "Sadly we won't be able to read their excellent solutions ;(
\nThey've deleted the accounts.
There are no real deletion in database, I am curious where their submission are comes from, maybe admin can tell.
", "votes": 2, "replies": [] } ] }, { "id": 1061423, "author_name": "sheep", "author_url": "", "post_date": "2020-10-27T02:21:39.430000", "content": "Now these 'talent' delete their account in order to avoid private share investigation.
\nI am highly recommand @philculliton to do similiary test between all submission. It's time to figure out where private share are comes from and tell us the source of their 'talent'.
\nI guess if this discussion not exist, they may finish their compeition 'normally'.
https://www.kaggle.com/c/rsna-str-pulmonary-embolism-detection/discussion/193338
\nThey have since deleted their GitHub account, with over 500 stars too.
", "votes": 2, "replies": [] }, { "id": 1061519, "author_name": "sin", "author_url": "", "post_date": "2020-10-27T03:59:29.420000", "content": "Also they changed their names :D. I am pretty sure @steamedsheep will get a solo gold if they were punished. Congrats for you solo gold finish!
", "votes": 4, "replies": [] }, { "id": 1062014, "author_name": "Ian Pan", "author_url": "", "post_date": "2020-10-27T14:00:16.960000", "content": "Suspicious to say the least. 2 top-10 solo competitors with potential ties to the 5-man 2nd place team who deleted their accounts at the conclusion of the competition.
\nIt seems like they were being used to submit dummy submissions perhaps at an attempt to get around the daily submission limit. Of course, this is all just speculation.
", "votes": 4, "replies": [] }, { "id": 1062024, "author_name": "Bo", "author_url": "", "post_date": "2020-10-27T14:08:20.487000", "content": "No. 13 team on public LB also deleted the account, in addition to No. 5 and No. 8 (cc @juliaelliott )
\nI also highly suspect these accounts were used by MSM team (public No. 1 team) to circumvent the daily submission limit. I noticed MSM was using \"fast submission\" throughout the competition. My guess is that they needed more subs in the end to debug \"real submission\" (even though CV and LB are highly correlated in this competition)
", "votes": 4, "replies": [] } ] }, { "id": 1060357, "author_name": "Salaryman", "author_url": "", "post_date": "2020-10-26T05:29:08.760000", "content": "I love and hate kaggle.
\nThe more I learn, the greater frustration I have, mostly by realizing the huge distance between the average (eg me) and the smart 😔😔
Achieving 0.15 after just 7 submissions.Five of them were submitted by the day before.It doesn't seem that he has worked hard. Otherwise, he will try to submit before, unless he is sure that he will get good results, so there is no need to submit before.Of course, it's just my guess…
", "votes": 1, "replies": [ { "id": 1060544, "author_name": "sheep", "author_url": "", "post_date": "2020-10-26T09:41:52.030000", "content": "Actually just 3 submissions, not 7.
", "votes": 2, "replies": [] } ] }, { "id": 1060661, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T12:55:14.123000", "content": "that Chinese guy in 1st place. He was alone for a while and then all of a sudden he had 4 more teammates. just about when the merging deadline was coming to a close. That was a bit suspicious.
\nEdit: why tf am I getting downvoted?
", "votes": -12, "replies": [ { "id": 1060686, "author_name": "Johnny Lee", "author_url": "", "post_date": "2020-10-26T13:18:33.290000", "content": "I don't know WHAT guy you are. But I do know why you're getting downvoted.
", "votes": 5, "replies": [] }, { "id": 1060711, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T13:48:58.290000", "content": "are you willing to tell me?
", "votes": -3, "replies": [] }, { "id": 1060732, "author_name": "Stanley Zheng", "author_url": "", "post_date": "2020-10-26T14:11:08.313000", "content": "Hey mate, noticed you getting downvoted quite a ton on this forum in general. It is generally better not to make assumptions and say things like \"TTA doesn't sound like it improves your score much\" or saying that someone is suspicious without evidence - newbie teams win competitions all the time, but true investigative journalism exists too, and Gilles' forum post is a great example. https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/175156
", "votes": 2, "replies": [] }, { "id": 1060762, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T14:32:37.677000", "content": "\n\nIt is generally better not to make assumptions and say things like saying that someone is suspicious without evidence
\n
if you have evidence then he's not a suspect anymore. He's been accused. I wasn't accusing anyone, I was just raising questions. Why would he choose to merge with just a week to go? Why are they all novices? And if they know each other why did they merge so late?
\n\n\nIt is generally better not to make assumptions and say things like \"TTA doesn't sound like it improves your score much\"
\n
Bro, it is generally accepted that hypothesizing is very much okay.
", "votes": -3, "replies": [] }, { "id": 1060877, "author_name": "", "author_url": "", "post_date": "2020-10-26T15:39:24.023000", "content": "", "votes": 0, "replies": [] }, { "id": 1060906, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T15:58:21.587000", "content": "I know, I'm just giving you my side of the story.
", "votes": -3, "replies": [] }, { "id": 1061251, "author_name": "sin", "author_url": "", "post_date": "2020-10-26T21:23:49.447000", "content": "How do you know he's Chinese??? If I put a gaggle as my profile picture, am I a gaggle???😑😑
", "votes": 2, "replies": [] }, { "id": 1061252, "author_name": "Stanley Zheng", "author_url": "", "post_date": "2020-10-26T21:25:11.807000", "content": "Haha @underwearfitting hilarious comment, never heard them being called \"gaggle\" but it's a perfect name.
\nIt also does say in his profile, \"Beijing China\"
@stanleyjzheng Kaggle's default pic is gaggle for a reason haha. Same logic here, if I put Tokyo on my profile, does that make me Japanese?😂
", "votes": 1, "replies": [] }, { "id": 1061263, "author_name": "DarkCube", "author_url": "", "post_date": "2020-10-26T21:49:03.607000", "content": "I mean I just said he was Chinese as a reference. There is no need to pull an investigation about it. Just as I'd say the one with the starry night a profile or the one with a baby.
", "votes": 0, "replies": [] } ] }, { "id": 1061694, "author_name": "SumanSudhir", "author_url": "", "post_date": "2020-10-27T08:08:37.113000", "content": "Hey @haqishen, can you please provide the link to the external data you have used. I am searching for additional PE dataset. [\"All Data Are Ext\"]
Sure, here you go
\nhttps://www.kaggle.com/c/deepfake-detection-challenge/discussion/157983