{
  "id": 254662,
  "title": "False Negative Rate Loss",
  "url": "/competitions/siim-covid19-detection/discussion/254662",
  "author_name": "",
  "post_date": "2021-07-23T05:10:07.466671800Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi all,</p>\n<p>I'm looking for a discussion/ paper that I saw on kaggle that mentioned weighting cross entropy loss with False Negative Rate. The idea being that with small FNRs algorithm already knows how to classify the class well so no need to focus on it. I've been trying to find the paper/ discussion about it with no luck. Would truly appreciate it if someone could link me to it.</p>\n<p>Thanks in advance :)</p>",
  "messages": [
    {
      "id": "1397336",
      "postDate": "07/23/2021 05:10:07",
      "content": "<p>Hi all,</p>\n<p>I'm looking for a discussion/ paper that I saw on kaggle that mentioned weighting cross entropy loss with False Negative Rate. The idea being that with small FNRs algorithm already knows how to classify the class well so no need to focus on it. I've been trying to find the paper/ discussion about it with no luck. Would truly appreciate it if someone could link me to it.</p>\n<p>Thanks in advance :)</p>",
      "rawMarkdown": "Hi all,\n\nI'm looking for a discussion/ paper that I saw on kaggle that mentioned weighting cross entropy loss with False Negative Rate. The idea being that with small FNRs algorithm already knows how to classify the class well so no need to focus on it. I've been trying to find the paper/ discussion about it with no luck. Would truly appreciate it if someone could link me to it.\n\nThanks in advance :)",
      "votes": null
    },
    {
      "id": "1400729",
      "postDate": "07/26/2021 14:36:34",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/sachin\" target=\"_blank\">@sachin</a>, you can have a look at this: <a href=\"http://tflearn.org/objectives/\" target=\"_blank\">http://tflearn.org/objectives/</a><br>\nThere is a loss function Weighted Crossentropy that implements what you are looking for</p>",
      "rawMarkdown": "Hi @sachin, you can have a look at this: http://tflearn.org/objectives/\nThere is a loss function Weighted Crossentropy that implements what you are looking for",
      "votes": null
    },
    {
      "id": "1400747",
      "postDate": "07/26/2021 14:51:31",
      "content": "<p>There is also a nice paper: <a href=\"https://arxiv.org/abs/2001.00570\" target=\"_blank\">https://arxiv.org/abs/2001.00570</a></p>",
      "rawMarkdown": "There is also a nice paper: https://arxiv.org/abs/2001.00570",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1400729,
      "author_name": "mikecho",
      "author_url": "",
      "post_date": "07/26/2021 14:36:34",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/sachin\" target=\"_blank\">@sachin</a>, you can have a look at this: <a href=\"http://tflearn.org/objectives/\" target=\"_blank\">http://tflearn.org/objectives/</a><br>\nThere is a loss function Weighted Crossentropy that implements what you are looking for</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1400747,
      "author_name": "mikecho",
      "author_url": "",
      "post_date": "07/26/2021 14:51:31",
      "content": "<p>There is also a nice paper: <a href=\"https://arxiv.org/abs/2001.00570\" target=\"_blank\">https://arxiv.org/abs/2001.00570</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1397336": "Hi all,\n\nI'm looking for a discussion/ paper that I saw on kaggle that mentioned weighting cross entropy loss with False Negative Rate. The idea being that with small FNRs algorithm already knows how to classify the class well so no need to focus on it. I've been trying to find the paper/ discussion about it with no luck. Would truly appreciate it if someone could link me to it.\n\nThanks in advance :)",
    "1400729": "Hi @sachin, you can have a look at this: http://tflearn.org/objectives/\nThere is a loss function Weighted Crossentropy that implements what you are looking for",
    "1400747": "There is also a nice paper: https://arxiv.org/abs/2001.00570"
  },
  "source": "meta"
}