{
  "id": 477680,
  "title": "Wavenet Skip Connection",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/477680",
  "author_name": "",
  "post_date": "2024-02-17T11:05:33.977886600Z",
  "votes": 7,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hi Kagglers! </p>\n<p>It's my first to study Deep Learning<br>\nSo, I have a Question😂</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16438831%2F713110a64a14e80e21e533737f289a8e%2Fw.PNG?generation=1708167823568663&amp;alt=media\"></p>\n<p>In chris notebook, he didn't use skip connection.<br>\nAnd i'm trying to add somethings. it cause running time longer.</p>\n<p><strong>So, i wanna to try skip connection, But i wonder Skipping connection in Wavenet is good option?</strong><br>\nFeel free to tell me</p>",
  "messages": [
    {
      "id": "2655971",
      "postDate": "02/17/2024 11:05:33",
      "content": "<p>Hi Kagglers! </p>\n<p>It's my first to study Deep Learning<br>\nSo, I have a Question😂</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16438831%2F713110a64a14e80e21e533737f289a8e%2Fw.PNG?generation=1708167823568663&amp;alt=media\"></p>\n<p>In chris notebook, he didn't use skip connection.<br>\nAnd i'm trying to add somethings. it cause running time longer.</p>\n<p><strong>So, i wanna to try skip connection, But i wonder Skipping connection in Wavenet is good option?</strong><br>\nFeel free to tell me</p>",
      "rawMarkdown": "Hi Kagglers! \n\nIt's my first to study Deep Learning\nSo, I have a Question😂\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16438831%2F713110a64a14e80e21e533737f289a8e%2Fw.PNG?generation=1708167823568663&alt=media)\n\nIn chris notebook, he didn't use skip connection.\nAnd i'm trying to add somethings. it cause running time longer.\n\n**So, i wanna to try skip connection, But i wonder Skipping connection in Wavenet is good option?**\nFeel free to tell me",
      "votes": null
    },
    {
      "id": "2655972",
      "postDate": "02/17/2024 11:07:59",
      "content": "<p>It's good, if the CV is improving :D</p>",
      "rawMarkdown": "It's good, if the CV is improving :D",
      "votes": null
    },
    {
      "id": "2656088",
      "postDate": "02/17/2024 12:49:28",
      "content": "<p>Thanks to telling me! if there're improvement, i'm gonna share my notebook</p>",
      "rawMarkdown": "Thanks to telling me! if there're improvement, i'm gonna share my notebook",
      "votes": null
    },
    {
      "id": "2656369",
      "postDate": "02/17/2024 16:38:27",
      "content": "<p>This logic applies to anything you'll try</p>",
      "rawMarkdown": "This logic applies to anything you'll try",
      "votes": null
    },
    {
      "id": "2659424",
      "postDate": "02/19/2024 20:22:43",
      "content": "<p><a href=\"https://www.kaggle.com/seoyunje\" target=\"_blank\">@seoyunje</a> As far as I know Residual and Skip connections are the same, Chris does have residual(skip) connections in his model</p>\n<p>From the chart, that looks like an auxiliary output. They were used back in early versions of inception, then dropped   later. Auxiliary outputs were abandoned due to their limited performance improvement and the additional computational overhead they introduced.</p>\n<p>I could be wrong!</p>",
      "rawMarkdown": "seoyunje As far as I know Residual and Skip connections are the same, Chris does have residual(skip) connections in his model\n\nFrom the chart, that looks like an auxiliary output. They were used back in early versions of inception, then dropped   later. Auxiliary outputs were abandoned due to their limited performance improvement and the additional computational overhead they introduced.\n\nI could be wrong!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2655972,
      "author_name": "martynoveduard",
      "author_url": "",
      "post_date": "02/17/2024 11:07:59",
      "content": "<p>It's good, if the CV is improving :D</p>",
      "votes": null,
      "replies": [
        {
          "id": 2656088,
          "author_name": "seoyunje",
          "author_url": "",
          "post_date": "02/17/2024 12:49:28",
          "content": "<p>Thanks to telling me! if there're improvement, i'm gonna share my notebook</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 2656369,
          "author_name": "gunesevitan",
          "author_url": "",
          "post_date": "02/17/2024 16:38:27",
          "content": "<p>This logic applies to anything you'll try</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2659424,
      "author_name": "nartaa",
      "author_url": "",
      "post_date": "02/19/2024 20:22:43",
      "content": "<p><a href=\"https://www.kaggle.com/seoyunje\" target=\"_blank\">@seoyunje</a> As far as I know Residual and Skip connections are the same, Chris does have residual(skip) connections in his model</p>\n<p>From the chart, that looks like an auxiliary output. They were used back in early versions of inception, then dropped   later. Auxiliary outputs were abandoned due to their limited performance improvement and the additional computational overhead they introduced.</p>\n<p>I could be wrong!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2655971": "Hi Kagglers! \n\nIt's my first to study Deep Learning\nSo, I have a Question😂\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16438831%2F713110a64a14e80e21e533737f289a8e%2Fw.PNG?generation=1708167823568663&alt=media)\n\nIn chris notebook, he didn't use skip connection.\nAnd i'm trying to add somethings. it cause running time longer.\n\n**So, i wanna to try skip connection, But i wonder Skipping connection in Wavenet is good option?**\nFeel free to tell me",
    "2655972": "It's good, if the CV is improving :D",
    "2656088": "Thanks to telling me! if there're improvement, i'm gonna share my notebook",
    "2656369": "This logic applies to anything you'll try",
    "2659424": "seoyunje As far as I know Residual and Skip connections are the same, Chris does have residual(skip) connections in his model\n\nFrom the chart, that looks like an auxiliary output. They were used back in early versions of inception, then dropped   later. Auxiliary outputs were abandoned due to their limited performance improvement and the additional computational overhead they introduced.\n\nI could be wrong!"
  },
  "source": "meta"
}