{
  "id": 399151,
  "title": "Fashion MNIST",
  "url": "/competitions/digit-recognizer/discussion/399151",
  "author_name": "Joshua Adrian Cahyono",
  "post_date": "2023-04-02T17:53:28.234000",
  "votes": 2,
  "comment_count": 1,
  "views": null,
  "content": "<p>I heard that now fashion MNIST has become the standard for algorithms, and some have said that it is better than MNIST. Question especially for more experienced data scientists in this field, what do you think? And also why does kaggle still use the original MNIST as a getting started competition.</p>",
  "messages": [
    {
      "id": 2254978,
      "postDate": "2023-05-11T11:57:55.777Z",
      "content": "<p>Hey <a href=\"https://www.kaggle.com/jvthunder\" target=\"_blank\">@jvthunder</a> <br>\nActually it's true MNIST is quite old like created in the 90's and fashion MNIST was created like some years ago in order to replace MNIST with a more \"complex\" dataset (accuracy on digit MNIST is quite awesome with \"too simple\" CNN or even SVM for example)<br>\nI guess as the MNIST is the first one Kaggle keeps it as an introduction… </p>",
      "rawMarkdown": "Hey @jvthunder \nActually it's true MNIST is quite old like created in the 90's and fashion MNIST was created like some years ago in order to replace MNIST with a more \"complex\" dataset (accuracy on digit MNIST is quite awesome with \"too simple\" CNN or even SVM for example)\nI guess as the MNIST is the first one Kaggle keeps it as an introduction... ",
      "votes": 2
    },
    {
      "id": 2206657,
      "postDate": "2023-04-02T17:53:28.233Z",
      "content": "<p>I heard that now fashion MNIST has become the standard for algorithms, and some have said that it is better than MNIST. Question especially for more experienced data scientists in this field, what do you think? And also why does kaggle still use the original MNIST as a getting started competition.</p>",
      "rawMarkdown": "I heard that now fashion MNIST has become the standard for algorithms, and some have said that it is better than MNIST. Question especially for more experienced data scientists in this field, what do you think? And also why does kaggle still use the original MNIST as a getting started competition.",
      "votes": 2
    }
  ],
  "comments": [
    {
      "id": 2254978,
      "author_name": "Antoine Gruson",
      "author_url": "",
      "post_date": "2023-05-11T11:57:55.777000",
      "content": "<p>Hey <a href=\"https://www.kaggle.com/jvthunder\" target=\"_blank\">@jvthunder</a> <br>\nActually it's true MNIST is quite old like created in the 90's and fashion MNIST was created like some years ago in order to replace MNIST with a more \"complex\" dataset (accuracy on digit MNIST is quite awesome with \"too simple\" CNN or even SVM for example)<br>\nI guess as the MNIST is the first one Kaggle keeps it as an introduction… </p>",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2254978": "Hey @jvthunder \nActually it's true MNIST is quite old like created in the 90's and fashion MNIST was created like some years ago in order to replace MNIST with a more \"complex\" dataset (accuracy on digit MNIST is quite awesome with \"too simple\" CNN or even SVM for example)\nI guess as the MNIST is the first one Kaggle keeps it as an introduction... ",
    "2206657": "I heard that now fashion MNIST has become the standard for algorithms, and some have said that it is better than MNIST. Question especially for more experienced data scientists in this field, what do you think? And also why does kaggle still use the original MNIST as a getting started competition."
  }
}