{
  "id": 48821,
  "title": "What features might have been learnt by CNN",
  "url": "/competitions/sp-society-camera-model-identification/discussion/48821",
  "author_name": "A0198918E_MaZhaoyang",
  "post_date": "2018-02-02T14:45:12.628000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Usually neural networks of CNN learn simple features to high-level features from initial layers to layers at the end. But in this case, what have been learnt should be something specific to cameras. It will be very interesting to know what have been learnt in each layer of CNN.</p>",
  "messages": [
    {
      "id": 277212,
      "postDate": "2018-02-02T14:45:12.630Z",
      "content": "<p>Usually neural networks of CNN learn simple features to high-level features from initial layers to layers at the end. But in this case, what have been learnt should be something specific to cameras. It will be very interesting to know what have been learnt in each layer of CNN.</p>",
      "rawMarkdown": "Usually neural networks of CNN learn simple features to high-level features from initial layers to layers at the end. But in this case, what have been learnt should be something specific to cameras. It will be very interesting to know what have been learnt in each layer of CNN."
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "277212": "Usually neural networks of CNN learn simple features to high-level features from initial layers to layers at the end. But in this case, what have been learnt should be something specific to cameras. It will be very interesting to know what have been learnt in each layer of CNN."
  }
}