{
  "id": 12629,
  "title": "Another prediction application",
  "url": "/competitions/datasciencebowl/discussion/12629",
  "author_name": "",
  "post_date": "2015-02-27T16:22:32.893Z",
  "votes": null,
  "comment_count": 2,
  "views": 634,
  "content": "<p>Dear Kagglers,</p>\n<p>&nbsp;Since I couldnt team up I think it is better to share software I have been working on. It is result of my Andrew NG coursera study. Works guite fast for small-middle size datasets. It is using batch gradient with Neural Networks, during my study I have also implemented Fmincg octave code in c++. Simply it runs a training session and it saves calculated weights on file system. You can set many parameters you want including loading previously saves weights. It has also multicore support. I couldnt get results I want since I couldnt handle my models well I think. As there is other or better alternatives this application maybe can help someone at some point. It should run mac or any linux distro without problem. There is no windows version yet. Since ML is not my main area I would also appreciate your feedbacks. Please find below github link, I believe instructions should be enough:</p>\n\n<p>https://github.com/buraksarac/NeuralNetwork</p>\n\n<p>Thank you</p>\n<p>Best Regards</p>\n<p>Burak</p>",
  "messages": [
    {
      "id": "65042",
      "postDate": "02/27/2015 16:22:32",
      "content": "<p>Dear Kagglers,</p>\n<p>&nbsp;Since I couldnt team up I think it is better to share software I have been working on. It is result of my Andrew NG coursera study. Works guite fast for small-middle size datasets. It is using batch gradient with Neural Networks, during my study I have also implemented Fmincg octave code in c++. Simply it runs a training session and it saves calculated weights on file system. You can set many parameters you want including loading previously saves weights. It has also multicore support. I couldnt get results I want since I couldnt handle my models well I think. As there is other or better alternatives this application maybe can help someone at some point. It should run mac or any linux distro without problem. There is no windows version yet. Since ML is not my main area I would also appreciate your feedbacks. Please find below github link, I believe instructions should be enough:</p>\n\n<p>https://github.com/buraksarac/NeuralNetwork</p>\n\n<p>Thank you</p>\n<p>Best Regards</p>\n<p>Burak</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "65053",
      "postDate": "02/27/2015 17:31:06",
      "content": "<p>@burak sarac &nbsp;Cool :) What are the advantages over caffe or cxxnet?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "65057",
      "postDate": "02/27/2015 18:29:35",
      "content": "<p>To be honest I dont know much about them:) Those projects are well community supported and developed in long term. I did this one in my way to study. Since it works I dont see any reason to not to share. I dont have any target to 'beat' any other project:)</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 65053,
      "author_name": "oldufo",
      "author_url": "",
      "post_date": "02/27/2015 17:31:06",
      "content": "<p>@burak sarac &nbsp;Cool :) What are the advantages over caffe or cxxnet?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 65057,
      "author_name": "hrgiger",
      "author_url": "",
      "post_date": "02/27/2015 18:29:35",
      "content": "<p>To be honest I dont know much about them:) Those projects are well community supported and developed in long term. I did this one in my way to study. Since it works I dont see any reason to not to share. I dont have any target to 'beat' any other project:)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "65042": "",
    "65053": "",
    "65057": ""
  },
  "source": "meta"
}