{
  "id": 203489,
  "title": "Deep Metric Learning ?",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/203489",
  "author_name": "",
  "post_date": "2020-12-15T14:05:38.547437800Z",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hi, I wonder if Deep Metric Learning can be used in this topic? If yes, can you guys give some tutorials.</p>",
  "messages": [
    {
      "id": "1113502",
      "postDate": "12/15/2020 14:05:38",
      "content": "<p>Hi, I wonder if Deep Metric Learning can be used in this topic? If yes, can you guys give some tutorials.</p>",
      "rawMarkdown": "Hi, I wonder if Deep Metric Learning can be used in this topic? If yes, can you guys give some tutorials.",
      "votes": null
    },
    {
      "id": "1113715",
      "postDate": "12/15/2020 16:34:58",
      "content": "<p>I tried this approach with a vgg16 architecture to extract 256 embedded features which I used in a SVM classifier. The obtained results where the same at around 0.81 (I got the same results with few layers trained on top of vgg16 arch.). Have not explored this approach further. Any other thoughts on how to go about this?</p>",
      "rawMarkdown": "I tried this approach with a vgg16 architecture to extract 256 embedded features which I used in a SVM classifier. The obtained results where the same at around 0.81 (I got the same results with few layers trained on top of vgg16 arch.). Have not explored this approach further. Any other thoughts on how to go about this?",
      "votes": null
    },
    {
      "id": "1116817",
      "postDate": "12/17/2020 13:41:28",
      "content": "<p>Can you define Deep Metric Learning in few sentences? and how would it help in this competition?</p>",
      "rawMarkdown": "Can you define Deep Metric Learning in few sentences? and how would it help in this competition?",
      "votes": null
    },
    {
      "id": "1116825",
      "postDate": "12/17/2020 13:45:49",
      "content": "<p>I haven't tried Metric Learning yet, but it has been used in other competitions,such as: <a href=\"https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109\" target=\"_blank\">https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109</a></p>",
      "rawMarkdown": "I haven't tried Metric Learning yet, but it has been used in other competitions,such as: https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109",
      "votes": null
    },
    {
      "id": "1116830",
      "postDate": "12/17/2020 13:51:11",
      "content": "<p>deep metric learning aims at learning similarity between samples using a distance metric. sort of feature extraction for images which can prove helpful in case of these noisy labels. <br>\nyou can refer to this <a href=\"https://towardsdatascience.com/deep-metric-learning-76fa0a5a415f\" target=\"_blank\">link</a> for a better understanding on this</p>",
      "rawMarkdown": "deep metric learning aims at learning similarity between samples using a distance metric. sort of feature extraction for images which can prove helpful in case of these noisy labels. \nyou can refer to this [link](https://towardsdatascience.com/deep-metric-learning-76fa0a5a415f) for a better understanding on this",
      "votes": null
    },
    {
      "id": "1116860",
      "postDate": "12/17/2020 14:21:04",
      "content": "<p>Thank you guys. Getting a deeper understanding of these different approaches so you can decide to use in your experiments is a hard task.</p>",
      "rawMarkdown": "Thank you guys. Getting a deeper understanding of these different approaches so you can decide to use in your experiments is a hard task.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1113715,
      "author_name": "kabhinay",
      "author_url": "",
      "post_date": "12/15/2020 16:34:58",
      "content": "<p>I tried this approach with a vgg16 architecture to extract 256 embedded features which I used in a SVM classifier. The obtained results where the same at around 0.81 (I got the same results with few layers trained on top of vgg16 arch.). Have not explored this approach further. Any other thoughts on how to go about this?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1116817,
      "author_name": "deepdreamx",
      "author_url": "",
      "post_date": "12/17/2020 13:41:28",
      "content": "<p>Can you define Deep Metric Learning in few sentences? and how would it help in this competition?</p>",
      "votes": null,
      "replies": [
        {
          "id": 1116825,
          "author_name": "thanhtinqn97",
          "author_url": "",
          "post_date": "12/17/2020 13:45:49",
          "content": "<p>I haven't tried Metric Learning yet, but it has been used in other competitions,such as: <a href=\"https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109\" target=\"_blank\">https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109</a></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1116830,
          "author_name": "kabhinay",
          "author_url": "",
          "post_date": "12/17/2020 13:51:11",
          "content": "<p>deep metric learning aims at learning similarity between samples using a distance metric. sort of feature extraction for images which can prove helpful in case of these noisy labels. <br>\nyou can refer to this <a href=\"https://towardsdatascience.com/deep-metric-learning-76fa0a5a415f\" target=\"_blank\">link</a> for a better understanding on this</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1116860,
          "author_name": "deepdreamx",
          "author_url": "",
          "post_date": "12/17/2020 14:21:04",
          "content": "<p>Thank you guys. Getting a deeper understanding of these different approaches so you can decide to use in your experiments is a hard task.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1113502": "Hi, I wonder if Deep Metric Learning can be used in this topic? If yes, can you guys give some tutorials.",
    "1113715": "I tried this approach with a vgg16 architecture to extract 256 embedded features which I used in a SVM classifier. The obtained results where the same at around 0.81 (I got the same results with few layers trained on top of vgg16 arch.). Have not explored this approach further. Any other thoughts on how to go about this?",
    "1116817": "Can you define Deep Metric Learning in few sentences? and how would it help in this competition?",
    "1116825": "I haven't tried Metric Learning yet, but it has been used in other competitions,such as: https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109",
    "1116830": "deep metric learning aims at learning similarity between samples using a distance metric. sort of feature extraction for images which can prove helpful in case of these noisy labels. \nyou can refer to this [link](https://towardsdatascience.com/deep-metric-learning-76fa0a5a415f) for a better understanding on this",
    "1116860": "Thank you guys. Getting a deeper understanding of these different approaches so you can decide to use in your experiments is a hard task."
  },
  "source": "meta"
}