{
  "id": 478810,
  "title": "[placeholder] experiment results ...",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/478810",
  "author_name": "",
  "post_date": "2024-02-22T08:06:35.531630100Z",
  "votes": 23,
  "comment_count": 3,
  "views": 0,
  "content": "<p>just got back from holidays …now start to work on kaggle.<br>\nto be updated ….</p>\n<p>i just do some quick literature survay and here are some initial plans:</p>\n<ul>\n<li>foundation model for EEG?</li>\n<li>Audio-based Joint-Embedding Predictive Architecture (A-JEPA) for self supervised learning? or I-JEP (image based)</li>\n<li>using diffusion for aux loss</li>\n<li>combination of graph NN and time series (or convert to circular scalp activation map?)</li>\n</ul>",
  "messages": [
    {
      "id": "2662913",
      "postDate": "02/22/2024 08:06:35",
      "content": "<p>just got back from holidays …now start to work on kaggle.<br>\nto be updated ….</p>\n<p>i just do some quick literature survay and here are some initial plans:</p>\n<ul>\n<li>foundation model for EEG?</li>\n<li>Audio-based Joint-Embedding Predictive Architecture (A-JEPA) for self supervised learning? or I-JEP (image based)</li>\n<li>using diffusion for aux loss</li>\n<li>combination of graph NN and time series (or convert to circular scalp activation map?)</li>\n</ul>",
      "rawMarkdown": "just got back from holidays ...now start to work on kaggle.\nto be updated ....\n\ni just do some quick literature survay and here are some initial plans:\n-  foundation model for EEG?\n-  Audio-based Joint-Embedding Predictive Architecture (A-JEPA) for self supervised learning? or I-JEP (image based)\n- using diffusion for aux loss\n- combination of graph NN and time series (or convert to circular scalp activation map?)",
      "votes": null
    },
    {
      "id": "2662945",
      "postDate": "02/22/2024 08:31:00",
      "content": "<p>Glad you and Chris are both in this competition</p>",
      "rawMarkdown": "Glad you and Chris are both in this competition",
      "votes": null
    },
    {
      "id": "2663366",
      "postDate": "02/22/2024 13:03:45",
      "content": "<p>Welcome <a href=\"https://www.kaggle.com/hengck23\" target=\"_blank\">@hengck23</a>! I saw some literature using graph nns as well, but I'm missing some understanding. I thought that GNNs in general did very poorly with regular graphs and the ones you could imagine from the double banana montage chains are highly regular. Very interested to see how this plays out!</p>",
      "rawMarkdown": "Welcome @hengck23! I saw some literature using graph nns as well, but I'm missing some understanding. I thought that GNNs in general did very poorly with regular graphs and the ones you could imagine from the double banana montage chains are highly regular. Very interested to see how this plays out!",
      "votes": null
    },
    {
      "id": "2696307",
      "postDate": "03/14/2024 08:23:07",
      "content": "<p>😭where？？？？</p>",
      "rawMarkdown": "😭where？？？？",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2662945,
      "author_name": "gentlezdh",
      "author_url": "",
      "post_date": "02/22/2024 08:31:00",
      "content": "<p>Glad you and Chris are both in this competition</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2663366,
      "author_name": "chemdatafarmer",
      "author_url": "",
      "post_date": "02/22/2024 13:03:45",
      "content": "<p>Welcome <a href=\"https://www.kaggle.com/hengck23\" target=\"_blank\">@hengck23</a>! I saw some literature using graph nns as well, but I'm missing some understanding. I thought that GNNs in general did very poorly with regular graphs and the ones you could imagine from the double banana montage chains are highly regular. Very interested to see how this plays out!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2696307,
      "author_name": "pkyangno1",
      "author_url": "",
      "post_date": "03/14/2024 08:23:07",
      "content": "<p>😭where？？？？</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2662913": "just got back from holidays ...now start to work on kaggle.\nto be updated ....\n\ni just do some quick literature survay and here are some initial plans:\n-  foundation model for EEG?\n-  Audio-based Joint-Embedding Predictive Architecture (A-JEPA) for self supervised learning? or I-JEP (image based)\n- using diffusion for aux loss\n- combination of graph NN and time series (or convert to circular scalp activation map?)",
    "2662945": "Glad you and Chris are both in this competition",
    "2663366": "Welcome @hengck23! I saw some literature using graph nns as well, but I'm missing some understanding. I thought that GNNs in general did very poorly with regular graphs and the ones you could imagine from the double banana montage chains are highly regular. Very interested to see how this plays out!",
    "2696307": "😭where？？？？"
  },
  "source": "meta"
}