{
  "id": 437303,
  "title": "OVERFITTING IN 'extravasation_injury' !",
  "url": "/competitions/rsna-2023-abdominal-trauma-detection/discussion/437303",
  "author_name": "justin1357",
  "post_date": "2023-09-06T08:40:28.928000",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>In the course of the experiment, I found that the probability predicted by the extravasation_injury' category increased 0.05(0.71-0.66), but the intuitive scaling of this small category would increase the risk of classification errors, and the classification error should be significantly supercharged, which would lead to a decline in accuracy, but in the lb. I suspect that the test concentration in lb contains more of the extravasation_injury' class, so it could lead to a serious shakeupgen by over-relying on lb to adjust the zoon factor, because if more of the extravasation_injury' is used in the public list, it will be less used for the private table.</p>",
  "messages": [
    {
      "id": 2425863,
      "postDate": "2023-09-06T08:40:28.930Z",
      "content": "<p>In the course of the experiment, I found that the probability predicted by the extravasation_injury' category increased 0.05(0.71-0.66), but the intuitive scaling of this small category would increase the risk of classification errors, and the classification error should be significantly supercharged, which would lead to a decline in accuracy, but in the lb. I suspect that the test concentration in lb contains more of the extravasation_injury' class, so it could lead to a serious shakeupgen by over-relying on lb to adjust the zoon factor, because if more of the extravasation_injury' is used in the public list, it will be less used for the private table.</p>",
      "rawMarkdown": "In the course of the experiment, I found that the probability predicted by the extravasation_injury' category increased 0.05(0.71-0.66), but the intuitive scaling of this small category would increase the risk of classification errors, and the classification error should be significantly supercharged, which would lead to a decline in accuracy, but in the lb. I suspect that the test concentration in lb contains more of the extravasation_injury' class, so it could lead to a serious shakeupgen by over-relying on lb to adjust the zoon factor, because if more of the extravasation_injury' is used in the public list, it will be less used for the private table.",
      "votes": 2
    },
    {
      "id": 2428814,
      "postDate": "2023-09-08T07:14:26.833Z",
      "content": "<p>Good observation</p>",
      "rawMarkdown": "Good observation"
    }
  ],
  "comments": [
    {
      "id": 2428814,
      "author_name": "t0xic-wast3",
      "author_url": "",
      "post_date": "2023-09-08T07:14:26.833000",
      "content": "<p>Good observation</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2425863": "In the course of the experiment, I found that the probability predicted by the extravasation_injury' category increased 0.05(0.71-0.66), but the intuitive scaling of this small category would increase the risk of classification errors, and the classification error should be significantly supercharged, which would lead to a decline in accuracy, but in the lb. I suspect that the test concentration in lb contains more of the extravasation_injury' class, so it could lead to a serious shakeupgen by over-relying on lb to adjust the zoon factor, because if more of the extravasation_injury' is used in the public list, it will be less used for the private table.",
    "2428814": "Good observation"
  }
}