{
  "id": 478745,
  "title": "The best single model",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/478745",
  "author_name": "",
  "post_date": "2024-02-22T01:42:23.371468700Z",
  "votes": 4,
  "comment_count": 5,
  "views": 0,
  "content": "<p>What is the VB/LB of your best single model</p>",
  "messages": [
    {
      "id": "2662560",
      "postDate": "02/22/2024 01:42:23",
      "content": "<p>What is the VB/LB of your best single model</p>",
      "rawMarkdown": "What is the VB/LB of your best single model",
      "votes": null
    },
    {
      "id": "2662620",
      "postDate": "02/22/2024 03:03:18",
      "content": "<p>EfficientNetB0 so far…</p>",
      "rawMarkdown": "EfficientNetB0 so far...",
      "votes": null
    },
    {
      "id": "2662644",
      "postDate": "02/22/2024 03:26:27",
      "content": "<p>I'm similar, I tried multi-modal approaches and the Transformer-based time series classification method was much worse than simple EfficientNetB0</p>",
      "rawMarkdown": "I'm similar, I tried multi-modal approaches and the Transformer-based time series classification method was much worse than simple EfficientNetB0",
      "votes": null
    },
    {
      "id": "2662649",
      "postDate": "02/22/2024 03:29:43",
      "content": "<p>i think time series will not work since the time framing of EEG spectrograms is not great enough to visualize time patterns, as for example, a sleeping brain, a stressed brain, etc…</p>",
      "rawMarkdown": "i think time series will not work since the time framing of EEG spectrograms is not great enough to visualize time patterns, as for example, a sleeping brain, a stressed brain, etc...",
      "votes": null
    },
    {
      "id": "2662656",
      "postDate": "02/22/2024 03:39:19",
      "content": "<p>Our best single model so far 0.33. </p>",
      "rawMarkdown": "Our best single model so far 0.33.",
      "votes": null
    },
    {
      "id": "2662663",
      "postDate": "02/22/2024 04:02:07",
      "content": "<p>I think if you wanna use time information, it would be better using eeg data not spectrogram</p>",
      "rawMarkdown": "I think if you wanna use time information, it would be better using eeg data not spectrogram",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2662620,
      "author_name": "gabrielfreddi",
      "author_url": "",
      "post_date": "02/22/2024 03:03:18",
      "content": "<p>EfficientNetB0 so far…</p>",
      "votes": null,
      "replies": [
        {
          "id": 2662644,
          "author_name": "chenboluo",
          "author_url": "",
          "post_date": "02/22/2024 03:26:27",
          "content": "<p>I'm similar, I tried multi-modal approaches and the Transformer-based time series classification method was much worse than simple EfficientNetB0</p>",
          "votes": null,
          "replies": [
            {
              "id": 2662649,
              "author_name": "gabrielfreddi",
              "author_url": "",
              "post_date": "02/22/2024 03:29:43",
              "content": "<p>i think time series will not work since the time framing of EEG spectrograms is not great enough to visualize time patterns, as for example, a sleeping brain, a stressed brain, etc…</p>",
              "votes": null,
              "replies": [
                {
                  "id": 2662663,
                  "author_name": "seoyunje",
                  "author_url": "",
                  "post_date": "02/22/2024 04:02:07",
                  "content": "<p>I think if you wanna use time information, it would be better using eeg data not spectrogram</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": 2662656,
      "author_name": "cody11null",
      "author_url": "",
      "post_date": "02/22/2024 03:39:19",
      "content": "<p>Our best single model so far 0.33. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2662560": "What is the VB/LB of your best single model",
    "2662620": "EfficientNetB0 so far...",
    "2662644": "I'm similar, I tried multi-modal approaches and the Transformer-based time series classification method was much worse than simple EfficientNetB0",
    "2662649": "i think time series will not work since the time framing of EEG spectrograms is not great enough to visualize time patterns, as for example, a sleeping brain, a stressed brain, etc...",
    "2662656": "Our best single model so far 0.33.",
    "2662663": "I think if you wanna use time information, it would be better using eeg data not spectrogram"
  },
  "source": "meta"
}