{
  "id": 12638,
  "title": "Looking to join with someone in top 60",
  "url": "/competitions/datasciencebowl/discussion/12638",
  "author_name": "",
  "post_date": "2015-02-28T15:06:53.987Z",
  "votes": null,
  "comment_count": 2,
  "views": 673,
  "content": "<p>Currently ranked #37 (0.705 public LB).</p>\n<p>My approach is mainly CNN using python (theano + lasagne), with a few others blended in as well.&nbsp;</p>\n<p>I've come up with one somewhat unorthodox method which appears&nbsp;to be&nbsp;helpful, but I have a feeling my limited experience designing and training CNN is a limiting factor.</p>\n<p>If you've been able to get&nbsp;a CNN to&nbsp;&lt;0.85 val loss (pre ensembling), then I think there's a decent&nbsp;chance we could&nbsp;help each other move up. PM me if you're interested and we can discuss more.</p>\n\n<p>P.S. I'd also like to work with someone on the Diabetic Retinopathy challenge after this competition is over, if you're interested in that as well this could be an&nbsp;opportunity to see if we work well together.</p>",
  "messages": [
    {
      "id": "65130",
      "postDate": "02/28/2015 15:06:53",
      "content": "<p>Currently ranked #37 (0.705 public LB).</p>\n<p>My approach is mainly CNN using python (theano + lasagne), with a few others blended in as well.&nbsp;</p>\n<p>I've come up with one somewhat unorthodox method which appears&nbsp;to be&nbsp;helpful, but I have a feeling my limited experience designing and training CNN is a limiting factor.</p>\n<p>If you've been able to get&nbsp;a CNN to&nbsp;&lt;0.85 val loss (pre ensembling), then I think there's a decent&nbsp;chance we could&nbsp;help each other move up. PM me if you're interested and we can discuss more.</p>\n\n<p>P.S. I'd also like to work with someone on the Diabetic Retinopathy challenge after this competition is over, if you're interested in that as well this could be an&nbsp;opportunity to see if we work well together.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "65140",
      "postDate": "02/28/2015 17:27:15",
      "content": "<p>I might be interested. I haven't done a competition in a while, so looks like I can't send private messages. Best single model in public LB is around .69. I'm using CNN with caffe. Message me if you're interested.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "65141",
      "postDate": "02/28/2015 17:39:57",
      "content": "<p>Hi emolson, I'm interested. I have used both cxx and&nbsp;theano for the competition. I have got 0.83 with a single model. I'm also interested in&nbsp;Diabetic Retinopathy challenge hope to start working on that after this one.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 65140,
      "author_name": "matthewkrump",
      "author_url": "",
      "post_date": "02/28/2015 17:27:15",
      "content": "<p>I might be interested. I haven't done a competition in a while, so looks like I can't send private messages. Best single model in public LB is around .69. I'm using CNN with caffe. Message me if you're interested.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 65141,
      "author_name": "akilaw",
      "author_url": "",
      "post_date": "02/28/2015 17:39:57",
      "content": "<p>Hi emolson, I'm interested. I have used both cxx and&nbsp;theano for the competition. I have got 0.83 with a single model. I'm also interested in&nbsp;Diabetic Retinopathy challenge hope to start working on that after this one.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "65130": "",
    "65140": "",
    "65141": ""
  },
  "source": "meta"
}