{
  "id": 281374,
  "title": "7th place solution",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/writeups/igor-lashkov-7th-place-solution",
  "author_name": "",
  "post_date": "2021-10-24T20:12:08.847Z",
  "votes": 8,
  "comment_count": 2,
  "views": 0,
  "content": "<p>First of all, I’d like to thank Kaggle, RSNA and the MICCAI Society for hosting this competition, as well has everyone here who shared great discussions and notebooks. For me, they were really valuable and knowledgeable! My congratulations to all the winners!</p>\n<p>Actually, I had time constraints, just one week to begin with this competition and make a submission. Given time limitations and provided competition dataset, initially I came up with a first easy solution of constructing a CNN model from scratch, no transfer learning was used here. There is no model ensembling, no big models, and no complex training techniques. No external datasets are used in this solution. either. </p>\n<p>I see the other competitors spent much time and effort to improve their solutions. There is a certain amount of luck I managed to get to the top. This is not my first experience of participating in Kaggle competitions, but rather first award for my previous efforts.</p>\n<p>The train/predict notebook is available here:<br>\n<a href=\"https://www.kaggle.com/igorlashkov/rsna-miccai-btumor-classification-finished\" target=\"_blank\">https://www.kaggle.com/igorlashkov/rsna-miccai-btumor-classification-finished</a></p>",
  "messages": [
    {
      "id": "1556308",
      "postDate": "10/24/2021 20:11:10",
      "content": "<p>First of all, I’d like to thank Kaggle, RSNA and the MICCAI Society for hosting this competition, as well has everyone here who shared great discussions and notebooks. For me, they were really valuable and knowledgeable! My congratulations to all the winners!</p>\n<p>Actually, I had time constraints, just one week to begin with this competition and make a submission. Given time limitations and provided competition dataset, initially I came up with a first easy solution of constructing a CNN model from scratch, no transfer learning was used here. There is no model ensembling, no big models, and no complex training techniques. No external datasets are used in this solution. either. </p>\n<p>I see the other competitors spent much time and effort to improve their solutions. There is a certain amount of luck I managed to get to the top. This is not my first experience of participating in Kaggle competitions, but rather first award for my previous efforts.</p>\n<p>The train/predict notebook is available here:<br>\n<a href=\"https://www.kaggle.com/igorlashkov/rsna-miccai-btumor-classification-finished\" target=\"_blank\">https://www.kaggle.com/igorlashkov/rsna-miccai-btumor-classification-finished</a></p>",
      "rawMarkdown": "First of all, I’d like to thank Kaggle, RSNA and the MICCAI Society for hosting this competition, as well has everyone here who shared great discussions and notebooks. For me, they were really valuable and knowledgeable! My congratulations to all the winners!\n\nActually, I had time constraints, just one week to begin with this competition and make a submission. Given time limitations and provided competition dataset, initially I came up with a first easy solution of constructing a CNN model from scratch, no transfer learning was used here. There is no model ensembling, no big models, and no complex training techniques. No external datasets are used in this solution. either. \n\nI see the other competitors spent much time and effort to improve their solutions. There is a certain amount of luck I managed to get to the top. This is not my first experience of participating in Kaggle competitions, but rather first award for my previous efforts.\n\nThe train/predict notebook is available here:\nhttps://www.kaggle.com/igorlashkov/rsna-miccai-btumor-classification-finished",
      "votes": null
    },
    {
      "id": "1556693",
      "postDate": "10/25/2021 05:18:16",
      "content": "<p><a href=\"https://www.kaggle.com/igorlashkov\" target=\"_blank\">@igorlashkov</a> Congrats!<br>\nOne week! That's one great combination of skills and luck.<br>\nI spent over 1.5 months on this competition and finally gave up after not seeing much progress, only to realize later that a couple more iterations MIGHT have landed me a bronze.</p>",
      "rawMarkdown": "igorlashkov Congrats!\nOne week! That's one great combination of skills and luck.\nI spent over 1.5 months on this competition and finally gave up after not seeing much progress, only to realize later that a couple more iterations MIGHT have landed me a bronze.",
      "votes": null
    },
    {
      "id": "1558915",
      "postDate": "10/26/2021 14:53:52",
      "content": "<p>Congratulations!</p>",
      "rawMarkdown": "Congratulations!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1556693,
      "author_name": "pranshu15",
      "author_url": "",
      "post_date": "10/25/2021 05:18:16",
      "content": "<p><a href=\"https://www.kaggle.com/igorlashkov\" target=\"_blank\">@igorlashkov</a> Congrats!<br>\nOne week! That's one great combination of skills and luck.<br>\nI spent over 1.5 months on this competition and finally gave up after not seeing much progress, only to realize later that a couple more iterations MIGHT have landed me a bronze.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1558915,
      "author_name": "atsunorifujita",
      "author_url": "",
      "post_date": "10/26/2021 14:53:52",
      "content": "<p>Congratulations!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1556308": "First of all, I’d like to thank Kaggle, RSNA and the MICCAI Society for hosting this competition, as well has everyone here who shared great discussions and notebooks. For me, they were really valuable and knowledgeable! My congratulations to all the winners!\n\nActually, I had time constraints, just one week to begin with this competition and make a submission. Given time limitations and provided competition dataset, initially I came up with a first easy solution of constructing a CNN model from scratch, no transfer learning was used here. There is no model ensembling, no big models, and no complex training techniques. No external datasets are used in this solution. either. \n\nI see the other competitors spent much time and effort to improve their solutions. There is a certain amount of luck I managed to get to the top. This is not my first experience of participating in Kaggle competitions, but rather first award for my previous efforts.\n\nThe train/predict notebook is available here:\nhttps://www.kaggle.com/igorlashkov/rsna-miccai-btumor-classification-finished",
    "1556693": "igorlashkov Congrats!\nOne week! That's one great combination of skills and luck.\nI spent over 1.5 months on this competition and finally gave up after not seeing much progress, only to realize later that a couple more iterations MIGHT have landed me a bronze.",
    "1558915": "Congratulations!"
  },
  "source": "meta"
}