{
  "id": 269626,
  "title": "3D CNN ideas",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/269626",
  "author_name": "Furkan K",
  "post_date": "2021-09-01T12:07:54.991000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi guys, since the top scoring NB uses efficientnet, we used that too. But it's obvious that model learn's nearly nothing. These might have some reasons. We though maybe it's because of the efficientnet network and we started to looking for other 3D cnns. But we couldn't find any 3D CNN that performs better than efficientnet.  Since other competitors might faced such a problem, this thread is for <code>well performing</code> 3D CNN links and their papers.</p>\n<p>Thanks.</p>",
  "messages": [
    {
      "id": 1499055,
      "postDate": "2021-09-01T12:07:54.993Z",
      "content": "<p>Hi guys, since the top scoring NB uses efficientnet, we used that too. But it's obvious that model learn's nearly nothing. These might have some reasons. We though maybe it's because of the efficientnet network and we started to looking for other 3D cnns. But we couldn't find any 3D CNN that performs better than efficientnet.  Since other competitors might faced such a problem, this thread is for <code>well performing</code> 3D CNN links and their papers.</p>\n<p>Thanks.</p>",
      "rawMarkdown": "Hi guys, since the top scoring NB uses efficientnet, we used that too. But it's obvious that model learn's nearly nothing. These might have some reasons. We though maybe it's because of the efficientnet network and we started to looking for other 3D cnns. But we couldn't find any 3D CNN that performs better than efficientnet.  Since other competitors might faced such a problem, this thread is for `well performing` 3D CNN links and their papers.\n\nThanks."
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1499055": "Hi guys, since the top scoring NB uses efficientnet, we used that too. But it's obvious that model learn's nearly nothing. These might have some reasons. We though maybe it's because of the efficientnet network and we started to looking for other 3D cnns. But we couldn't find any 3D CNN that performs better than efficientnet.  Since other competitors might faced such a problem, this thread is for `well performing` 3D CNN links and their papers.\n\nThanks."
  }
}