{
  "id": 97935,
  "title": "Kaggle GANs",
  "url": "/competitions/generative-dog-images/discussion/97935",
  "author_name": "JM100",
  "post_date": "2019-06-29T23:45:30.038000",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I think this is the first competition of GANs in Kaggle. This talk: <a href=\"https://www.youtube.com/watch?v=o1EbDcrUTEk&amp;list=PLqFaTIg4myu99Huiyr2ZojzN3dlYhe8rf&amp;index=2&amp;t=0s\">Kaggle Days San Francisco</a>\npresent the vision of Kaggle toward GAN competitions.</p>",
  "messages": [
    {
      "id": 564753,
      "postDate": "2019-06-29T23:45:30.037Z",
      "content": "<p>I think this is the first competition of GANs in Kaggle. This talk: <a href=\"https://www.youtube.com/watch?v=o1EbDcrUTEk&amp;list=PLqFaTIg4myu99Huiyr2ZojzN3dlYhe8rf&amp;index=2&amp;t=0s\">Kaggle Days San Francisco</a>\npresent the vision of Kaggle toward GAN competitions.</p>",
      "rawMarkdown": "I think this is the first competition of GANs in Kaggle. This talk: [Kaggle Days San Francisco](https://www.youtube.com/watch?v=o1EbDcrUTEk&amp;list=PLqFaTIg4myu99Huiyr2ZojzN3dlYhe8rf&amp;index=2&amp;t=0s)\npresent the vision of Kaggle toward GAN competitions.",
      "votes": 2
    },
    {
      "id": 564770,
      "postDate": "2019-06-30T00:37:12.607Z",
      "rawMarkdown": "",
      "isDeleted": true,
      "replies": [
        {
          "id": 564778,
          "postDate": "2019-06-30T01:08:23.973Z",
          "content": "<p>However, PATE-GANs, which uses precisely the method you described to achieve differential privacy, still performs poorly when augmenting data for training other tasks, as demonstrated in the original paper.\n<a href=\"https://openreview.net/forum?id=S1zk9iRqF7\">https://openreview.net/forum?id=S1zk9iRqF7</a></p>",
          "rawMarkdown": "However, PATE-GANs, which uses precisely the method you described to achieve differential privacy, still performs poorly when augmenting data for training other tasks, as demonstrated in the original paper.\nhttps://openreview.net/forum?id=S1zk9iRqF7",
          "votes": 2
        }
      ]
    },
    {
      "id": 565008,
      "postDate": "2019-06-30T09:39:19.770Z",
      "content": "<p>Thanks for sharing </p>",
      "rawMarkdown": "Thanks for sharing "
    }
  ],
  "comments": [
    {
      "id": 564770,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-06-30T00:37:12.607000",
      "content": "",
      "votes": 0,
      "replies": [
        {
          "id": 564778,
          "author_name": "Ching-Yuan Bai",
          "author_url": "",
          "post_date": "2019-06-30T01:08:23.973000",
          "content": "<p>However, PATE-GANs, which uses precisely the method you described to achieve differential privacy, still performs poorly when augmenting data for training other tasks, as demonstrated in the original paper.\n<a href=\"https://openreview.net/forum?id=S1zk9iRqF7\">https://openreview.net/forum?id=S1zk9iRqF7</a></p>",
          "votes": 2,
          "replies": []
        }
      ]
    },
    {
      "id": 565008,
      "author_name": "Madhur Sabherwal",
      "author_url": "",
      "post_date": "2019-06-30T09:39:19.770000",
      "content": "<p>Thanks for sharing </p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "564753": "I think this is the first competition of GANs in Kaggle. This talk: [Kaggle Days San Francisco](https://www.youtube.com/watch?v=o1EbDcrUTEk&amp;list=PLqFaTIg4myu99Huiyr2ZojzN3dlYhe8rf&amp;index=2&amp;t=0s)\npresent the vision of Kaggle toward GAN competitions.",
    "564770": "",
    "565008": "Thanks for sharing "
  }
}