{
  "id": 546073,
  "title": "What works in coughing classification? ",
  "url": "/competitions/airs-ai-in-respiratory-sounds/discussion/546073",
  "author_name": "",
  "post_date": "2024-11-13T18:59:11.572044800Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Curios to know what works in this competition. </p>\n<p>In our case audio was the worst, HeAR embeddings as features in the middle (about 0.65) and tabular data with demographic information brought most of the result.</p>\n<p>Have someone tried OPERA fine-tuning? </p>",
  "messages": [
    {
      "id": "3044736",
      "postDate": "11/13/2024 18:59:11",
      "content": "<p>Curios to know what works in this competition. </p>\n<p>In our case audio was the worst, HeAR embeddings as features in the middle (about 0.65) and tabular data with demographic information brought most of the result.</p>\n<p>Have someone tried OPERA fine-tuning? </p>",
      "rawMarkdown": "Curios to know what works in this competition. \n\nIn our case audio was the worst, HeAR embeddings as features in the middle (about 0.65) and tabular data with demographic information brought most of the result.\n\nHave someone tried OPERA fine-tuning?",
      "votes": null
    },
    {
      "id": "3047243",
      "postDate": "11/16/2024 12:40:38",
      "content": "<p><a href=\"https://www.kaggle.com/samvelkoch\" target=\"_blank\">@samvelkoch</a> I was also hoping to see the results with opera. It would have been really interesting to compare it with HeAR. do you want to work with me on this? I am thinking of adding those embeddings as well in the dataset.</p>",
      "rawMarkdown": "samvelkoch I was also hoping to see the results with opera. It would have been really interesting to compare it with HeAR. do you want to work with me on this? I am thinking of adding those embeddings as well in the dataset.",
      "votes": null
    },
    {
      "id": "3047352",
      "postDate": "11/16/2024 15:30:23",
      "content": "<p><a href=\"https://www.kaggle.com/jeevahimanshukaushik\" target=\"_blank\">@jeevahimanshukaushik</a> let's have a talk. We are able to continue on this issue. Will send you PM</p>",
      "rawMarkdown": "jeevahimanshukaushik let's have a talk. We are able to continue on this issue. Will send you PM",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3047243,
      "author_name": "jeevahimanshukaushik",
      "author_url": "",
      "post_date": "11/16/2024 12:40:38",
      "content": "<p><a href=\"https://www.kaggle.com/samvelkoch\" target=\"_blank\">@samvelkoch</a> I was also hoping to see the results with opera. It would have been really interesting to compare it with HeAR. do you want to work with me on this? I am thinking of adding those embeddings as well in the dataset.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3047352,
          "author_name": "samvelkoch",
          "author_url": "",
          "post_date": "11/16/2024 15:30:23",
          "content": "<p><a href=\"https://www.kaggle.com/jeevahimanshukaushik\" target=\"_blank\">@jeevahimanshukaushik</a> let's have a talk. We are able to continue on this issue. Will send you PM</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3044736": "Curios to know what works in this competition. \n\nIn our case audio was the worst, HeAR embeddings as features in the middle (about 0.65) and tabular data with demographic information brought most of the result.\n\nHave someone tried OPERA fine-tuning?",
    "3047243": "samvelkoch I was also hoping to see the results with opera. It would have been really interesting to compare it with HeAR. do you want to work with me on this? I am thinking of adding those embeddings as well in the dataset.",
    "3047352": "jeevahimanshukaushik let's have a talk. We are able to continue on this issue. Will send you PM"
  },
  "source": "meta"
}