{
  "id": 347984,
  "title": "How to find the optimal threshold for each organ?",
  "url": "/competitions/hubmap-organ-segmentation/discussion/347984",
  "author_name": "",
  "post_date": "2022-08-26T07:44:25.798961300Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello folks, I am using transformer-based approach for this task using resized data (768,768). I was wondering how should I find what is the optimal Threshold for my model for each of the organs?</p>\n<p>Any advice would be really helpful.<br>\nThanks! :)</p>",
  "messages": [
    {
      "id": "1914578",
      "postDate": "08/26/2022 07:44:25",
      "content": "<p>Hello folks, I am using transformer-based approach for this task using resized data (768,768). I was wondering how should I find what is the optimal Threshold for my model for each of the organs?</p>\n<p>Any advice would be really helpful.<br>\nThanks! :)</p>",
      "rawMarkdown": "Hello folks, I am using transformer-based approach for this task using resized data (768,768). I was wondering how should I find what is the optimal Threshold for my model for each of the organs?\n\nAny advice would be really helpful.\nThanks! :)",
      "votes": null
    },
    {
      "id": "1915096",
      "postDate": "08/26/2022 16:52:52",
      "content": "<ol>\n<li>Save OOF predictions.</li>\n<li>Tune thresholds with them maximizing or minimizing metrics.</li>\n</ol>",
      "rawMarkdown": "1. Save OOF predictions.\n2. Tune thresholds with them maximizing or minimizing metrics.",
      "votes": null
    },
    {
      "id": "1915572",
      "postDate": "08/27/2022 05:08:57",
      "content": "<p>That's the way but it's not that simple.</p>\n<ul>\n<li>Visualize ground-truth labels and understand the noise</li>\n<li>Tweak thresholds and visualize your predictions</li>\n<li>Try to understand how can you capture the noise</li>\n<li>Decide whether you should capture the noise or not</li>\n</ul>",
      "rawMarkdown": "That's the way but it's not that simple.\n\n- Visualize ground-truth labels and understand the noise\n- Tweak thresholds and visualize your predictions\n- Try to understand how can you capture the noise\n- Decide whether you should capture the noise or not",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1915096,
      "author_name": "vad13irt",
      "author_url": "",
      "post_date": "08/26/2022 16:52:52",
      "content": "<ol>\n<li>Save OOF predictions.</li>\n<li>Tune thresholds with them maximizing or minimizing metrics.</li>\n</ol>",
      "votes": null,
      "replies": [
        {
          "id": 1915572,
          "author_name": "gunesevitan",
          "author_url": "",
          "post_date": "08/27/2022 05:08:57",
          "content": "<p>That's the way but it's not that simple.</p>\n<ul>\n<li>Visualize ground-truth labels and understand the noise</li>\n<li>Tweak thresholds and visualize your predictions</li>\n<li>Try to understand how can you capture the noise</li>\n<li>Decide whether you should capture the noise or not</li>\n</ul>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1914578": "Hello folks, I am using transformer-based approach for this task using resized data (768,768). I was wondering how should I find what is the optimal Threshold for my model for each of the organs?\n\nAny advice would be really helpful.\nThanks! :)",
    "1915096": "1. Save OOF predictions.\n2. Tune thresholds with them maximizing or minimizing metrics.",
    "1915572": "That's the way but it's not that simple.\n\n- Visualize ground-truth labels and understand the noise\n- Tweak thresholds and visualize your predictions\n- Try to understand how can you capture the noise\n- Decide whether you should capture the noise or not"
  },
  "source": "meta"
}