{
  "id": 418639,
  "title": "Standardization for distance based algorithms for image pixel data",
  "url": "/competitions/digit-recognizer/discussion/418639",
  "author_name": "Sai pavan",
  "post_date": "2023-06-21T19:42:10.959000",
  "votes": 2,
  "comment_count": 0,
  "views": null,
  "content": "<p>While solving the problem, i have found some interesting things about whether should we apply standardization of distance based algorithms like KNN,especially for images</p>\n<p>Usually,It is recommended to standardize the data when using K-Nearest Neighbors (KNN) algorithm. Standardization helps to bring all features to a similar scale, which is important for distance-based algorithms like KNN.</p>\n<p>But, In image classification tasks, the spatial relationship and distribution of pixel values are crucial for capturing patterns and features. Applying standardization to pixel values would flatten the distribution and distort the original characteristics of the images, potentially harming the performance of the KNN algorithm..</p>",
  "messages": [
    {
      "id": 2312284,
      "postDate": "2023-06-21T19:42:10.960Z",
      "content": "<p>While solving the problem, i have found some interesting things about whether should we apply standardization of distance based algorithms like KNN,especially for images</p>\n<p>Usually,It is recommended to standardize the data when using K-Nearest Neighbors (KNN) algorithm. Standardization helps to bring all features to a similar scale, which is important for distance-based algorithms like KNN.</p>\n<p>But, In image classification tasks, the spatial relationship and distribution of pixel values are crucial for capturing patterns and features. Applying standardization to pixel values would flatten the distribution and distort the original characteristics of the images, potentially harming the performance of the KNN algorithm..</p>",
      "rawMarkdown": "While solving the problem, i have found some interesting things about whether should we apply standardization of distance based algorithms like KNN,especially for images\n\nUsually,It is recommended to standardize the data when using K-Nearest Neighbors (KNN) algorithm. Standardization helps to bring all features to a similar scale, which is important for distance-based algorithms like KNN.\n\nBut, In image classification tasks, the spatial relationship and distribution of pixel values are crucial for capturing patterns and features. Applying standardization to pixel values would flatten the distribution and distort the original characteristics of the images, potentially harming the performance of the KNN algorithm..\n",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2312284": "While solving the problem, i have found some interesting things about whether should we apply standardization of distance based algorithms like KNN,especially for images\n\nUsually,It is recommended to standardize the data when using K-Nearest Neighbors (KNN) algorithm. Standardization helps to bring all features to a similar scale, which is important for distance-based algorithms like KNN.\n\nBut, In image classification tasks, the spatial relationship and distribution of pixel values are crucial for capturing patterns and features. Applying standardization to pixel values would flatten the distribution and distort the original characteristics of the images, potentially harming the performance of the KNN algorithm..\n"
  }
}