{
  "id": 19119,
  "title": "Swedish Institute of Computer Science solution",
  "url": "/competitions/seizure-prediction/discussion/19119",
  "author_name": "Theodore Vasiloudis",
  "post_date": "2016-02-22T14:54:14.523000",
  "votes": 4,
  "comment_count": 0,
  "views": 1124,
  "content": "<p>Hello all,</p>\n\n<p>better late than never we are releasing the code for our solution, which\ngot us a top-30 finish.</p>\n\n<p>A few specific contributions we made might prove useful;\nlike new CV rules for sklearn that allowed us to split segments in\na way that reduced over-fitting and our parallelized data input/output\nmethods.</p>\n\n<p>You can find the code here: <a href=\"https://github.com/sics-lm/kaggle-seizure-prediction\">https://github.com/sics-lm/kaggle-seizure-prediction</a></p>\n\n<p>Regards, <br>\nTheodore</p>",
  "messages": [
    {
      "id": 109012,
      "postDate": "2016-02-22T14:54:14.523Z",
      "content": "<p>Hello all,</p>\n\n<p>better late than never we are releasing the code for our solution, which\ngot us a top-30 finish.</p>\n\n<p>A few specific contributions we made might prove useful;\nlike new CV rules for sklearn that allowed us to split segments in\na way that reduced over-fitting and our parallelized data input/output\nmethods.</p>\n\n<p>You can find the code here: <a href=\"https://github.com/sics-lm/kaggle-seizure-prediction\">https://github.com/sics-lm/kaggle-seizure-prediction</a></p>\n\n<p>Regards, <br>\nTheodore</p>",
      "rawMarkdown": "Hello all,\r\n\r\nbetter late than never we are releasing the code for our solution, which\r\ngot us a top-30 finish.\r\n\r\nA few specific contributions we made might prove useful;\r\nlike new CV rules for sklearn that allowed us to split segments in\r\na way that reduced over-fitting and our parallelized data input/output\r\nmethods.\r\n\r\nYou can find the code here: https://github.com/sics-lm/kaggle-seizure-prediction\r\n\r\nRegards,  \r\nTheodore",
      "votes": 4
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "109012": "Hello all,\r\n\r\nbetter late than never we are releasing the code for our solution, which\r\ngot us a top-30 finish.\r\n\r\nA few specific contributions we made might prove useful;\r\nlike new CV rules for sklearn that allowed us to split segments in\r\na way that reduced over-fitting and our parallelized data input/output\r\nmethods.\r\n\r\nYou can find the code here: https://github.com/sics-lm/kaggle-seizure-prediction\r\n\r\nRegards,  \r\nTheodore"
  }
}