{
  "id": 272261,
  "title": "Beginner's Approach",
  "url": "/competitions/digit-recognizer/discussion/272261",
  "author_name": "Elemento",
  "post_date": "2021-09-14T18:16:59.576000",
  "votes": 3,
  "comment_count": 0,
  "views": null,
  "content": "<p>Hey guys, it's one of my early competitions on Kaggle after the Titanic one, and I would love to share my work. In this <a href=\"https://www.kaggle.com/elemento/digitrecognizer-1\" target=\"_blank\">notebook</a>, </p>\n<ul>\n<li>The key features of this notebook are <strong>Scaling the pixel values</strong>, and <strong>Data Augmentation</strong>.</li>\n<li>As for the architectures, I tried out various different ones, using Conv2D, Dense, Dropout, MaxPool2D, BatchNormalization, etc.</li>\n<li>I experimented with various optimizers such as RMSProp, Adam, SGD, etc, and different callbacks like Reduce Learning Rate on Plateau and Early Stop.</li>\n</ul>\n<p>I would love to get your suggestions. Please do leave them in the comments section. Happy Kaggling 😁</p>",
  "messages": [
    {
      "id": 1513006,
      "postDate": "2021-09-14T18:16:59.577Z",
      "content": "<p>Hey guys, it's one of my early competitions on Kaggle after the Titanic one, and I would love to share my work. In this <a href=\"https://www.kaggle.com/elemento/digitrecognizer-1\" target=\"_blank\">notebook</a>, </p>\n<ul>\n<li>The key features of this notebook are <strong>Scaling the pixel values</strong>, and <strong>Data Augmentation</strong>.</li>\n<li>As for the architectures, I tried out various different ones, using Conv2D, Dense, Dropout, MaxPool2D, BatchNormalization, etc.</li>\n<li>I experimented with various optimizers such as RMSProp, Adam, SGD, etc, and different callbacks like Reduce Learning Rate on Plateau and Early Stop.</li>\n</ul>\n<p>I would love to get your suggestions. Please do leave them in the comments section. Happy Kaggling 😁</p>",
      "rawMarkdown": "Hey guys, it's one of my early competitions on Kaggle after the Titanic one, and I would love to share my work. In this [notebook](https://www.kaggle.com/elemento/digitrecognizer-1), \n- The key features of this notebook are **Scaling the pixel values**, and **Data Augmentation**.\n- As for the architectures, I tried out various different ones, using Conv2D, Dense, Dropout, MaxPool2D, BatchNormalization, etc.\n- I experimented with various optimizers such as RMSProp, Adam, SGD, etc, and different callbacks like Reduce Learning Rate on Plateau and Early Stop.\n\nI would love to get your suggestions. Please do leave them in the comments section. Happy Kaggling 😁",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1513006": "Hey guys, it's one of my early competitions on Kaggle after the Titanic one, and I would love to share my work. In this [notebook](https://www.kaggle.com/elemento/digitrecognizer-1), \n- The key features of this notebook are **Scaling the pixel values**, and **Data Augmentation**.\n- As for the architectures, I tried out various different ones, using Conv2D, Dense, Dropout, MaxPool2D, BatchNormalization, etc.\n- I experimented with various optimizers such as RMSProp, Adam, SGD, etc, and different callbacks like Reduce Learning Rate on Plateau and Early Stop.\n\nI would love to get your suggestions. Please do leave them in the comments section. Happy Kaggling 😁"
  }
}