{
  "id": 168914,
  "title": "Scale in the images: Is it a leak?",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/168914",
  "author_name": "",
  "post_date": "2020-07-22T09:59:50.203221900Z",
  "votes": 3,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I observed that most images with a scale in them are malignant. Did anyone observe this? \nOr I would say malignant images seem to have a higher probability of having a scale within them. </p>",
  "messages": [
    {
      "id": "939569",
      "postDate": "07/22/2020 09:59:50",
      "content": "<p>I observed that most images with a scale in them are malignant. Did anyone observe this? \nOr I would say malignant images seem to have a higher probability of having a scale within them. </p>",
      "rawMarkdown": "I observed that most images with a scale in them are malignant. Did anyone observe this? \nOr I would say malignant images seem to have a higher probability of having a scale within them.",
      "votes": null
    },
    {
      "id": "939631",
      "postDate": "07/22/2020 10:46:25",
      "content": "<p>Not true, there are many Benign with a scale, blue markings, black markings / black sheet, microscope, etc.</p>\n\n<p>Just to verify, could you please post your observations i.e. the number of Malignant images with scale vs total number of images with scale.</p>",
      "rawMarkdown": "Not true, there are many Benign with a scale, blue markings, black markings / black sheet, microscope, etc.\n\nJust to verify, could you please post your observations i.e. the number of Malignant images with scale vs total number of images with scale.",
      "votes": null
    },
    {
      "id": "939699",
      "postDate": "07/22/2020 11:44:06",
      "content": "<p>That was just a casual observation by randomly looking at malignant images. Do not have any such stats yet. </p>",
      "rawMarkdown": "That was just a casual observation by randomly looking at malignant images. Do not have any such stats yet.",
      "votes": null
    },
    {
      "id": "945377",
      "postDate": "07/25/2020 19:09:11",
      "content": "<p>This is worth looking into. If I remember correctly, there's a story (possibly anecdotal) about a model that achieved a stunning accuracy on a difficult classification problem - only to realize that all of the pictures of one class were taken when the sky was cloudy and the model simply learned a brightness filter.</p>",
      "rawMarkdown": "This is worth looking into. If I remember correctly, there's a story (possibly anecdotal) about a model that achieved a stunning accuracy on a difficult classification problem - only to realize that all of the pictures of one class were taken when the sky was cloudy and the model simply learned a brightness filter.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 939631,
      "author_name": "sirishks",
      "author_url": "",
      "post_date": "07/22/2020 10:46:25",
      "content": "<p>Not true, there are many Benign with a scale, blue markings, black markings / black sheet, microscope, etc.</p>\n\n<p>Just to verify, could you please post your observations i.e. the number of Malignant images with scale vs total number of images with scale.</p>",
      "votes": null,
      "replies": [
        {
          "id": 939699,
          "author_name": "krisho007",
          "author_url": "",
          "post_date": "07/22/2020 11:44:06",
          "content": "<p>That was just a casual observation by randomly looking at malignant images. Do not have any such stats yet. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 945377,
      "author_name": "thebayou",
      "author_url": "",
      "post_date": "07/25/2020 19:09:11",
      "content": "<p>This is worth looking into. If I remember correctly, there's a story (possibly anecdotal) about a model that achieved a stunning accuracy on a difficult classification problem - only to realize that all of the pictures of one class were taken when the sky was cloudy and the model simply learned a brightness filter.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "939569": "I observed that most images with a scale in them are malignant. Did anyone observe this? \nOr I would say malignant images seem to have a higher probability of having a scale within them.",
    "939631": "Not true, there are many Benign with a scale, blue markings, black markings / black sheet, microscope, etc.\n\nJust to verify, could you please post your observations i.e. the number of Malignant images with scale vs total number of images with scale.",
    "939699": "That was just a casual observation by randomly looking at malignant images. Do not have any such stats yet.",
    "945377": "This is worth looking into. If I remember correctly, there's a story (possibly anecdotal) about a model that achieved a stunning accuracy on a difficult classification problem - only to realize that all of the pictures of one class were taken when the sky was cloudy and the model simply learned a brightness filter."
  },
  "source": "meta"
}