{
  "id": 45515,
  "title": "Hard to classify 'on' vs 'off'",
  "url": "/competitions/tensorflow-speech-recognition-challenge/discussion/45515",
  "author_name": "Ildoo Kim",
  "post_date": "2017-12-12T13:14:25.223000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I saw confusion matrix from my model, </p>\n\n<p>error rate between class on and class off is much higher.</p>\n\n<p>Does Anyone use a specific technique to deal with this kind of situation?</p>\n\n<p>I guess I can change data preprocessing(currently, Spectrogram), or use Data resampling. But I guess there are many other options.</p>",
  "messages": [
    {
      "id": 256642,
      "postDate": "2017-12-12T13:14:25.223Z",
      "content": "<p>I saw confusion matrix from my model, </p>\n\n<p>error rate between class on and class off is much higher.</p>\n\n<p>Does Anyone use a specific technique to deal with this kind of situation?</p>\n\n<p>I guess I can change data preprocessing(currently, Spectrogram), or use Data resampling. But I guess there are many other options.</p>",
      "rawMarkdown": "I saw confusion matrix from my model, \n\nerror rate between class on and class off is much higher.\n\nDoes Anyone use a specific technique to deal with this kind of situation?\n\nI guess I can change data preprocessing(currently, Spectrogram), or use Data resampling. But I guess there are many other options."
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "256642": "I saw confusion matrix from my model, \n\nerror rate between class on and class off is much higher.\n\nDoes Anyone use a specific technique to deal with this kind of situation?\n\nI guess I can change data preprocessing(currently, Spectrogram), or use Data resampling. But I guess there are many other options."
  }
}