{
  "id": 13189,
  "title": "Dear (future) top 10 participants",
  "url": "/competitions/malware-classification/discussion/13189",
  "author_name": "",
  "post_date": "2015-04-01T18:08:28.203Z",
  "votes": 1,
  "comment_count": 11,
  "views": 2904,
  "content": "<p>Could you tell me what to do now? I'm basically just eye-balling each malware now : (</p>",
  "messages": [
    {
      "id": "69348",
      "postDate": "04/01/2015 18:08:28",
      "content": "<p>Could you tell me what to do now? I'm basically just eye-balling each malware now : (</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69351",
      "postDate": "04/01/2015 18:33:32",
      "content": "<p>Try not to overfit? All scores in top 10 are quite close to each other and we are tested on only on 3 K cases. It may be that you are first in the private leaderboard and you just don't know it!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69352",
      "postDate": "04/01/2015 18:35:39",
      "content": "<p>rcarson, could you tell me what to do now?</p>\n<p>You are in top 10 =D</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69378",
      "postDate": "04/02/2015 01:43:10",
      "content": "<p>I would tell you everything, but I am only Top 11 at the moment :(</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69382",
      "postDate": "04/02/2015 01:53:42",
      "content": "<p>[quote=NxGTR;69378]</p>\n<p>I would tell you everything, but I am only Top 11 at the moment :(</p>\n<p>[/quote]</p>\n<p>LOL!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69395",
      "postDate": "04/02/2015 03:52:54",
      "content": "<p>[quote=NxGTR;69378]</p>\n<p>I would tell you everything, but I am only Top 11 at the moment :(</p>\n<p>[/quote]</p>\n<p>You have my upvote!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69396",
      "postDate": "04/02/2015 03:54:04",
      "content": "<p>[quote=Mikhail Trofimov;69352]</p>\n<p>rcarson, could you tell me what to do now?</p>\n<p>You are in top 10 =D</p>\n<p>[/quote]</p>\n<p>I decide to do more feature engineering. There should be something more in 500 GB data :D</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69397",
      "postDate": "04/02/2015 03:55:51",
      "content": "<p>[quote=&#924;&#945;&#961;&#953;&#959;&#962; &#924;&#953;&#967;&#945;&#951;&#955;&#953;&#948;&#951;&#962; KazAnova;69351]</p>\n<p>Try not to overfit? All scores in top 10 are quite close to each other and we are tested on only on 3 K cases. It may be that you are first in the private leaderboard and you just don't know it!</p>\n<p>[/quote]</p>\n<p>Actually I think the contrary. I believe my model has an outlier that is terribly wrong and it is in private LB data. Why? Because my hand calibration doesn't work :D</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69569",
      "postDate": "04/03/2015 21:30:58",
      "content": "<p>Is there any trick i'm missing to incorporate 2-grams using&nbsp;a MacPro with 4GB RAM?</p>\n<p>There are 2^16 = 65536 of them...</p>\n<p>=p</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69574",
      "postDate": "04/03/2015 22:14:40",
      "content": "<p>Maybe&nbsp;you can extract 2grams from a small batch first, like 1000 samples,&nbsp;and&nbsp;discard the 2 g-rams that are useless. I think you will end up with much fewer 2 grams than 65536.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "72221",
      "postDate": "04/18/2015 01:05:28",
      "content": "<p>@rcarson, now you know what to do!</p>\n<p>@KazAnova, you were right! =D</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "72251",
      "postDate": "04/18/2015 08:07:58",
      "content": "<p>[quote=Mikhail Trofimov;72221]</p>\n<p>@rcarson, now you know what to do!</p>\n<p>@KazAnova, you were right! =D</p>\n<p>[/quote]</p>\n<p>:D</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 69351,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "04/01/2015 18:33:32",
      "content": "<p>Try not to overfit? All scores in top 10 are quite close to each other and we are tested on only on 3 K cases. It may be that you are first in the private leaderboard and you just don't know it!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69352,
      "author_name": "mikhailtrofimov",
      "author_url": "",
      "post_date": "04/01/2015 18:35:39",
      "content": "<p>rcarson, could you tell me what to do now?</p>\n<p>You are in top 10 =D</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69378,
      "author_name": "carloshuertas",
      "author_url": "",
      "post_date": "04/02/2015 01:43:10",
      "content": "<p>I would tell you everything, but I am only Top 11 at the moment :(</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69382,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "04/02/2015 01:53:42",
      "content": "<p>[quote=NxGTR;69378]</p>\n<p>I would tell you everything, but I am only Top 11 at the moment :(</p>\n<p>[/quote]</p>\n<p>LOL!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69395,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "04/02/2015 03:52:54",
      "content": "<p>[quote=NxGTR;69378]</p>\n<p>I would tell you everything, but I am only Top 11 at the moment :(</p>\n<p>[/quote]</p>\n<p>You have my upvote!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69396,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "04/02/2015 03:54:04",
      "content": "<p>[quote=Mikhail Trofimov;69352]</p>\n<p>rcarson, could you tell me what to do now?</p>\n<p>You are in top 10 =D</p>\n<p>[/quote]</p>\n<p>I decide to do more feature engineering. There should be something more in 500 GB data :D</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69397,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "04/02/2015 03:55:51",
      "content": "<p>[quote=&#924;&#945;&#961;&#953;&#959;&#962; &#924;&#953;&#967;&#945;&#951;&#955;&#953;&#948;&#951;&#962; KazAnova;69351]</p>\n<p>Try not to overfit? All scores in top 10 are quite close to each other and we are tested on only on 3 K cases. It may be that you are first in the private leaderboard and you just don't know it!</p>\n<p>[/quote]</p>\n<p>Actually I think the contrary. I believe my model has an outlier that is terribly wrong and it is in private LB data. Why? Because my hand calibration doesn't work :D</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69569,
      "author_name": "snowdog",
      "author_url": "",
      "post_date": "04/03/2015 21:30:58",
      "content": "<p>Is there any trick i'm missing to incorporate 2-grams using&nbsp;a MacPro with 4GB RAM?</p>\n<p>There are 2^16 = 65536 of them...</p>\n<p>=p</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69574,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "04/03/2015 22:14:40",
      "content": "<p>Maybe&nbsp;you can extract 2grams from a small batch first, like 1000 samples,&nbsp;and&nbsp;discard the 2 g-rams that are useless. I think you will end up with much fewer 2 grams than 65536.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 72221,
      "author_name": "mikhailtrofimov",
      "author_url": "",
      "post_date": "04/18/2015 01:05:28",
      "content": "<p>@rcarson, now you know what to do!</p>\n<p>@KazAnova, you were right! =D</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 72251,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "04/18/2015 08:07:58",
      "content": "<p>[quote=Mikhail Trofimov;72221]</p>\n<p>@rcarson, now you know what to do!</p>\n<p>@KazAnova, you were right! =D</p>\n<p>[/quote]</p>\n<p>:D</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "69348": "",
    "69351": "",
    "69352": "",
    "69378": "",
    "69382": "",
    "69395": "",
    "69396": "",
    "69397": "",
    "69569": "",
    "69574": "",
    "72221": "",
    "72251": ""
  },
  "source": "meta"
}