{
  "id": 485206,
  "title": "Any success in data augmentations?",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/485206",
  "author_name": "",
  "post_date": "2024-03-19T21:35:22.874530200Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hey guys,</p>\n<p>I've tried multiple data augmentation techniques but had mostly no success. </p>\n<p>I tried the XYMasking newly added to albumentation, but it only slightly decreased my lb score. </p>\n<p>What are the data augmentation methods you are using and are they successful?</p>\n<p>My best performing model in a single-model two stage training. any other tricks only decreased the performance.</p>",
  "messages": [
    {
      "id": "2706424",
      "postDate": "03/19/2024 21:35:22",
      "content": "<p>Hey guys,</p>\n<p>I've tried multiple data augmentation techniques but had mostly no success. </p>\n<p>I tried the XYMasking newly added to albumentation, but it only slightly decreased my lb score. </p>\n<p>What are the data augmentation methods you are using and are they successful?</p>\n<p>My best performing model in a single-model two stage training. any other tricks only decreased the performance.</p>",
      "rawMarkdown": "Hey guys,\n\nI've tried multiple data augmentation techniques but had mostly no success. \n\nI tried the XYMasking newly added to albumentation, but it only slightly decreased my lb score. \n\nWhat are the data augmentation methods you are using and are they successful?\n\nMy best performing model in a single-model two stage training. any other tricks only decreased the performance.",
      "votes": null
    },
    {
      "id": "2706805",
      "postDate": "03/20/2024 05:17:56",
      "content": "<p>How about augmentation just only 50s spectrograms(not 10m)?<br>\nIn my case, it was good option</p>",
      "rawMarkdown": "How about augmentation just only 50s spectrograms(not 10m)?\nIn my case, it was good option",
      "votes": null
    },
    {
      "id": "2716606",
      "postDate": "03/26/2024 05:58:29",
      "content": "<p>mixup can increase cv and lb a little bit</p>",
      "rawMarkdown": "mixup can increase cv and lb a little bit",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2706805,
      "author_name": "seoyunje",
      "author_url": "",
      "post_date": "03/20/2024 05:17:56",
      "content": "<p>How about augmentation just only 50s spectrograms(not 10m)?<br>\nIn my case, it was good option</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2716606,
      "author_name": "jimmyisme1",
      "author_url": "",
      "post_date": "03/26/2024 05:58:29",
      "content": "<p>mixup can increase cv and lb a little bit</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2706424": "Hey guys,\n\nI've tried multiple data augmentation techniques but had mostly no success. \n\nI tried the XYMasking newly added to albumentation, but it only slightly decreased my lb score. \n\nWhat are the data augmentation methods you are using and are they successful?\n\nMy best performing model in a single-model two stage training. any other tricks only decreased the performance.",
    "2706805": "How about augmentation just only 50s spectrograms(not 10m)?\nIn my case, it was good option",
    "2716606": "mixup can increase cv and lb a little bit"
  },
  "source": "meta"
}