{
  "id": 240294,
  "title": "Exact Weighted Log Loss Function to compare results of RSNA Challenge to another dataset ",
  "url": "/competitions/rsna-intracranial-hemorrhage-detection/discussion/240294",
  "author_name": "RebekkaGörge",
  "post_date": "2021-05-19T07:19:40.585000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hello, <br>\nI am currently writing my masterthesis in the field of deep learning to detect intracranial hemorrhage detection. Therefore, I am trying to compare the results of my model trained on the ICH to another dataset. But as the test ground truth labels are not available, I need to be sure that I compare the results on both datasets with the exact same weigthed mean log loss. But I am not sure if the weights here are [2,1,1,1,1,1] or for example [2/7, 1/7, 1/7, 1/7, 1/7, 1/7]. <br>\nIs there any way to find out the exact weights that are used in the second test submission? </p>",
  "messages": [
    {
      "id": 1314455,
      "postDate": "2021-05-19T07:19:40.587Z",
      "content": "<p>Hello, <br>\nI am currently writing my masterthesis in the field of deep learning to detect intracranial hemorrhage detection. Therefore, I am trying to compare the results of my model trained on the ICH to another dataset. But as the test ground truth labels are not available, I need to be sure that I compare the results on both datasets with the exact same weigthed mean log loss. But I am not sure if the weights here are [2,1,1,1,1,1] or for example [2/7, 1/7, 1/7, 1/7, 1/7, 1/7]. <br>\nIs there any way to find out the exact weights that are used in the second test submission? </p>",
      "rawMarkdown": "Hello, \nI am currently writing my masterthesis in the field of deep learning to detect intracranial hemorrhage detection. Therefore, I am trying to compare the results of my model trained on the ICH to another dataset. But as the test ground truth labels are not available, I need to be sure that I compare the results on both datasets with the exact same weigthed mean log loss. But I am not sure if the weights here are [2,1,1,1,1,1] or for example [2/7, 1/7, 1/7, 1/7, 1/7, 1/7]. \nIs there any way to find out the exact weights that are used in the second test submission? "
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1314455": "Hello, \nI am currently writing my masterthesis in the field of deep learning to detect intracranial hemorrhage detection. Therefore, I am trying to compare the results of my model trained on the ICH to another dataset. But as the test ground truth labels are not available, I need to be sure that I compare the results on both datasets with the exact same weigthed mean log loss. But I am not sure if the weights here are [2,1,1,1,1,1] or for example [2/7, 1/7, 1/7, 1/7, 1/7, 1/7]. \nIs there any way to find out the exact weights that are used in the second test submission? "
  }
}