{
  "id": 250578,
  "title": "useful for beginner: 🐜 how to debug algorithm (not software)",
  "url": "/competitions/siim-covid19-detection/discussion/250578",
  "author_name": "hengck23",
  "post_date": "2021-07-03T12:07:32.065000",
  "votes": 44,
  "comment_count": 7,
  "views": 0,
  "content": "<p><img src=\"https://i.ibb.co/KWGwThP/Selection-492.png\" alt=\"https://i.ibb.co/KWGwThP/Selection-492.png\"><br>\n<a href=\"https://fullstackdeeplearning.com/spring2021/lecture-7/\" target=\"_blank\">https://fullstackdeeplearning.com/spring2021/lecture-7/</a><br>\nTroubleshooting Deep Neural Networks | Full Stack Deep Learning | Spring 2021 | Lecture 7 🐜</p>\n<p>algorithm bugs are difficult to catch. 🐜 they behave as noise. they inhibit  accuracy and in some very rare case, they could improve results (good bug … little noise is good regularisation)</p>\n<p>🐜 🐜 </p>",
  "messages": [
    {
      "id": 1374553,
      "postDate": "2021-07-03T12:07:32.067Z",
      "content": "<p><img src=\"https://i.ibb.co/KWGwThP/Selection-492.png\" alt=\"https://i.ibb.co/KWGwThP/Selection-492.png\"><br>\n<a href=\"https://fullstackdeeplearning.com/spring2021/lecture-7/\" target=\"_blank\">https://fullstackdeeplearning.com/spring2021/lecture-7/</a><br>\nTroubleshooting Deep Neural Networks | Full Stack Deep Learning | Spring 2021 | Lecture 7 🐜</p>\n<p>algorithm bugs are difficult to catch. 🐜 they behave as noise. they inhibit  accuracy and in some very rare case, they could improve results (good bug … little noise is good regularisation)</p>\n<p>🐜 🐜 </p>",
      "rawMarkdown": "![https://i.ibb.co/KWGwThP/Selection-492.png](https://i.ibb.co/KWGwThP/Selection-492.png)\nhttps://fullstackdeeplearning.com/spring2021/lecture-7/\nTroubleshooting Deep Neural Networks | Full Stack Deep Learning | Spring 2021 | Lecture 7 🐜\n\nalgorithm bugs are difficult to catch. 🐜 they behave as noise. they inhibit  accuracy and in some very rare case, they could improve results (good bug ... little noise is good regularisation)\n\n🐜 🐜 ",
      "votes": 44
    },
    {
      "id": 1374570,
      "postDate": "2021-07-03T12:16:35.643Z",
      "content": "<p><img src=\"https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png\" alt=\"https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png\"></p>",
      "rawMarkdown": "![https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png](https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png)",
      "votes": 10,
      "replies": [
        {
          "id": 1374609,
          "postDate": "2021-07-03T12:58:05.253Z",
          "content": "<p>Thanks for sharing. </p>",
          "rawMarkdown": "Thanks for sharing. "
        }
      ]
    },
    {
      "id": 1375211,
      "postDate": "2021-07-04T02:56:47.433Z",
      "content": "<p><a href=\"https://www.youtube.com/channel/UC2of8xmN_238SaqJB3ryv0w/videos\" target=\"_blank\">https://www.youtube.com/channel/UC2of8xmN_238SaqJB3ryv0w/videos</a><br>\nLearning with Limited and Imperfect Data</p>\n<p>data are imperfect. what should you do next?<br>\nThese youtube videos explain the very few papers that explain WHY it would work.<br>\ne.g. <br>\nPhillip Isola -- When and Why Does Contrastive Learing Work?<br>\nBoqing Gong -- When Vision Transformers Outperform ResNets without Pretraining or Strong Data Augmentations?</p>",
      "rawMarkdown": "https://www.youtube.com/channel/UC2of8xmN_238SaqJB3ryv0w/videos\nLearning with Limited and Imperfect Data\n\ndata are imperfect. what should you do next?\nThese youtube videos explain the very few papers that explain WHY it would work.\ne.g. \nPhillip Isola -- When and Why Does Contrastive Learing Work?\nBoqing Gong -- When Vision Transformers Outperform ResNets without Pretraining or Strong Data Augmentations?\n",
      "votes": 2
    },
    {
      "id": 1376743,
      "postDate": "2021-07-05T09:52:03.233Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 1376756,
      "postDate": "2021-07-05T09:58:39.713Z",
      "content": "<p>Nice, thanks for share</p>",
      "rawMarkdown": "Nice, thanks for share"
    },
    {
      "id": 1376669,
      "postDate": "2021-07-05T09:00:38.313Z",
      "content": "<p>Thanks for sharing!</p>",
      "rawMarkdown": "Thanks for sharing!"
    },
    {
      "id": 1375228,
      "postDate": "2021-07-04T03:32:15.963Z",
      "content": "<p>Thanks for sharing…</p>",
      "rawMarkdown": "Thanks for sharing..."
    }
  ],
  "comments": [
    {
      "id": 1374570,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "2021-07-03T12:16:35.643000",
      "content": "<p><img src=\"https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png\" alt=\"https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png\"></p>",
      "votes": 10,
      "replies": [
        {
          "id": 1374609,
          "author_name": "Samarth Gupta",
          "author_url": "",
          "post_date": "2021-07-03T12:58:05.253000",
          "content": "<p>Thanks for sharing. </p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1375211,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "2021-07-04T02:56:47.433000",
      "content": "<p><a href=\"https://www.youtube.com/channel/UC2of8xmN_238SaqJB3ryv0w/videos\" target=\"_blank\">https://www.youtube.com/channel/UC2of8xmN_238SaqJB3ryv0w/videos</a><br>\nLearning with Limited and Imperfect Data</p>\n<p>data are imperfect. what should you do next?<br>\nThese youtube videos explain the very few papers that explain WHY it would work.<br>\ne.g. <br>\nPhillip Isola -- When and Why Does Contrastive Learing Work?<br>\nBoqing Gong -- When Vision Transformers Outperform ResNets without Pretraining or Strong Data Augmentations?</p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 1376743,
      "author_name": "",
      "author_url": "",
      "post_date": "2021-07-05T09:52:03.233000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 1376756,
      "author_name": "sharmina es",
      "author_url": "",
      "post_date": "2021-07-05T09:58:39.713000",
      "content": "<p>Nice, thanks for share</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 1376669,
      "author_name": "Deni Gelaev",
      "author_url": "",
      "post_date": "2021-07-05T09:00:38.313000",
      "content": "<p>Thanks for sharing!</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 1375228,
      "author_name": "laxman kusuma",
      "author_url": "",
      "post_date": "2021-07-04T03:32:15.963000",
      "content": "<p>Thanks for sharing…</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1374553": "![https://i.ibb.co/KWGwThP/Selection-492.png](https://i.ibb.co/KWGwThP/Selection-492.png)\nhttps://fullstackdeeplearning.com/spring2021/lecture-7/\nTroubleshooting Deep Neural Networks | Full Stack Deep Learning | Spring 2021 | Lecture 7 🐜\n\nalgorithm bugs are difficult to catch. 🐜 they behave as noise. they inhibit  accuracy and in some very rare case, they could improve results (good bug ... little noise is good regularisation)\n\n🐜 🐜 ",
    "1374570": "![https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png](https://fullstackdeeplearning.com/spring2021/lecture-7-notes-media/image15.png)",
    "1375211": "https://www.youtube.com/channel/UC2of8xmN_238SaqJB3ryv0w/videos\nLearning with Limited and Imperfect Data\n\ndata are imperfect. what should you do next?\nThese youtube videos explain the very few papers that explain WHY it would work.\ne.g. \nPhillip Isola -- When and Why Does Contrastive Learing Work?\nBoqing Gong -- When Vision Transformers Outperform ResNets without Pretraining or Strong Data Augmentations?\n",
    "1376743": "",
    "1376756": "Nice, thanks for share",
    "1376669": "Thanks for sharing!",
    "1375228": "Thanks for sharing..."
  }
}