{
  "id": 56217,
  "title": "Now Best way is to recover LB 12hours ago",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/56217",
  "author_name": "",
  "post_date": "2018-05-07T22:22:50.243881200Z",
  "votes": 6,
  "comment_count": 9,
  "views": 0,
  "content": "<p>Since lots of people blend with that kernel or just simply lower the learning rate, the result submission must be different, it's too difficult to track. Most efficient way will be just recover the leader board 12 hours ago before that kernel published.</p>\n\n<p>Downvote is normal, because of lots of. copy,run,summit guy, they aren't willing to let this happen.</p>",
  "messages": [
    {
      "id": "324690",
      "postDate": "05/07/2018 22:22:50",
      "content": "<p>Since lots of people blend with that kernel or just simply lower the learning rate, the result submission must be different, it's too difficult to track. Most efficient way will be just recover the leader board 12 hours ago before that kernel published.</p>\n\n<p>Downvote is normal, because of lots of. copy,run,summit guy, they aren't willing to let this happen.</p>",
      "rawMarkdown": "Since lots of people blend with that kernel or just simply lower the learning rate, the result submission must be different, it's too difficult to track. Most efficient way will be just recover the leader board 12 hours ago before that kernel published.\n\nDownvote is normal, because of lots of. copy,run,summit guy, they aren't willing to let this happen.",
      "votes": null
    },
    {
      "id": "324707",
      "postDate": "05/07/2018 22:39:32",
      "content": "<p>Nice one :)</p>",
      "rawMarkdown": "Nice one :)",
      "votes": null
    },
    {
      "id": "324710",
      "postDate": "05/07/2018 22:43:28",
      "content": "<p>Call this guy <img src=\"https://i.imgur.com/6MMWajS.jpg\" alt=\"LB 12h back\"></p>",
      "rawMarkdown": "Call this guy ![LB 12h back][1]\n\n\n  [1]: https://i.imgur.com/6MMWajS.jpg",
      "votes": null
    },
    {
      "id": "324711",
      "postDate": "05/07/2018 22:44:34",
      "content": "<p>That may sound a good idea but in some competitions , some people/teams wait until the last hours or even the last minutes (winner of instacart ) to submit their best  model.  And this, independantly of released kernels.</p>",
      "rawMarkdown": "That may sound a good idea but in some competitions , some people/teams wait until the last hours or even the last minutes (winner of instacart ) to submit their best  model.  And this, independantly of released kernels.",
      "votes": null
    },
    {
      "id": "324718",
      "postDate": "05/07/2018 22:49:16",
      "content": "<p>Maybe ask for help from the organizer to choose the submitted files before that kernel uploaded.</p>",
      "rawMarkdown": "Maybe ask for help from the organizer to choose the submitted files before that kernel uploaded.",
      "votes": null
    },
    {
      "id": "324729",
      "postDate": "05/07/2018 23:01:16",
      "content": "<p>Undestand your pain Wenjie Bai as this kernel shuffled considerably the leaderboard... But let us wait for the final score. I expect a lot of movements around....  And as said, some people wait until the last minute... exactly like for exam at school :-)</p>",
      "rawMarkdown": "Undestand your pain Wenjie Bai as this kernel shuffled considerably the leaderboard... But let us wait for the final score. I expect a lot of movements around....  And as said, some people wait until the last minute... exactly like for exam at school :-)",
      "votes": null
    },
    {
      "id": "324932",
      "postDate": "05/08/2018 01:52:07",
      "content": "<p>nice idea, although didn't resolve the concern completely given they're still able to submit the file after LB recover.</p>",
      "rawMarkdown": "nice idea, although didn't resolve the concern completely given they're still able to submit the file after LB recover.",
      "votes": null
    },
    {
      "id": "324989",
      "postDate": "05/08/2018 02:46:36",
      "content": "<p>I agree with you</p>",
      "rawMarkdown": "I agree with you",
      "votes": null
    },
    {
      "id": "325037",
      "postDate": "05/08/2018 03:45:40",
      "content": "<p>I am doing competition weekly on <a href=\"https://numer.ai/\">Numerai</a>, they use correlation to determine the uniqueness of each submission (I may be wrong). Kaggle may utilize a similar strategy to prevent copy and paste solution and/or blend of blend of ... of blend of public kernels' submissions for future competitions, for example Avito Demand Prediction Challenge.</p>",
      "rawMarkdown": "I am doing competition weekly on [Numerai][1], they use correlation to determine the uniqueness of each submission (I may be wrong). Kaggle may utilize a similar strategy to prevent copy and paste solution and/or blend of blend of ... of blend of public kernels' submissions for future competitions, for example Avito Demand Prediction Challenge.\n\n\n  [1]: https://numer.ai/",
      "votes": null
    },
    {
      "id": "357576",
      "postDate": "07/16/2018 10:53:21",
      "content": "<p>You are absolutely right...!!</p>",
      "rawMarkdown": "You are absolutely right...!!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 324707,
      "author_name": "",
      "author_url": "",
      "post_date": "05/07/2018 22:39:32",
      "content": "<p>Nice one :)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 324710,
      "author_name": "atanasova",
      "author_url": "",
      "post_date": "05/07/2018 22:43:28",
      "content": "<p>Call this guy <img src=\"https://i.imgur.com/6MMWajS.jpg\" alt=\"LB 12h back\"></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 324711,
      "author_name": "serigne",
      "author_url": "",
      "post_date": "05/07/2018 22:44:34",
      "content": "<p>That may sound a good idea but in some competitions , some people/teams wait until the last hours or even the last minutes (winner of instacart ) to submit their best  model.  And this, independantly of released kernels.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 324718,
      "author_name": "creamiracle",
      "author_url": "",
      "post_date": "05/07/2018 22:49:16",
      "content": "<p>Maybe ask for help from the organizer to choose the submitted files before that kernel uploaded.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 324729,
      "author_name": "ericbenhamou",
      "author_url": "",
      "post_date": "05/07/2018 23:01:16",
      "content": "<p>Undestand your pain Wenjie Bai as this kernel shuffled considerably the leaderboard... But let us wait for the final score. I expect a lot of movements around....  And as said, some people wait until the last minute... exactly like for exam at school :-)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 324932,
      "author_name": "dongxu027",
      "author_url": "",
      "post_date": "05/08/2018 01:52:07",
      "content": "<p>nice idea, although didn't resolve the concern completely given they're still able to submit the file after LB recover.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 324989,
      "author_name": "loleno",
      "author_url": "",
      "post_date": "05/08/2018 02:46:36",
      "content": "<p>I agree with you</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 325037,
      "author_name": "konohayui",
      "author_url": "",
      "post_date": "05/08/2018 03:45:40",
      "content": "<p>I am doing competition weekly on <a href=\"https://numer.ai/\">Numerai</a>, they use correlation to determine the uniqueness of each submission (I may be wrong). Kaggle may utilize a similar strategy to prevent copy and paste solution and/or blend of blend of ... of blend of public kernels' submissions for future competitions, for example Avito Demand Prediction Challenge.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 357576,
      "author_name": "ashishpatel26",
      "author_url": "",
      "post_date": "07/16/2018 10:53:21",
      "content": "<p>You are absolutely right...!!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "324690": "Since lots of people blend with that kernel or just simply lower the learning rate, the result submission must be different, it's too difficult to track. Most efficient way will be just recover the leader board 12 hours ago before that kernel published.\n\nDownvote is normal, because of lots of. copy,run,summit guy, they aren't willing to let this happen.",
    "324707": "Nice one :)",
    "324710": "Call this guy ![LB 12h back][1]\n\n\n  [1]: https://i.imgur.com/6MMWajS.jpg",
    "324711": "That may sound a good idea but in some competitions , some people/teams wait until the last hours or even the last minutes (winner of instacart ) to submit their best  model.  And this, independantly of released kernels.",
    "324718": "Maybe ask for help from the organizer to choose the submitted files before that kernel uploaded.",
    "324729": "Undestand your pain Wenjie Bai as this kernel shuffled considerably the leaderboard... But let us wait for the final score. I expect a lot of movements around....  And as said, some people wait until the last minute... exactly like for exam at school :-)",
    "324932": "nice idea, although didn't resolve the concern completely given they're still able to submit the file after LB recover.",
    "324989": "I agree with you",
    "325037": "I am doing competition weekly on [Numerai][1], they use correlation to determine the uniqueness of each submission (I may be wrong). Kaggle may utilize a similar strategy to prevent copy and paste solution and/or blend of blend of ... of blend of public kernels' submissions for future competitions, for example Avito Demand Prediction Challenge.\n\n\n  [1]: https://numer.ai/",
    "357576": "You are absolutely right...!!"
  },
  "source": "meta"
}