{
  "id": 28321,
  "title": "Streaming Data/Online learning (Example)",
  "url": "/competitions/dogs-vs-cats-redux-kernels-edition/discussion/28321",
  "author_name": "Ronin",
  "post_date": "2017-01-31T23:30:13.448000",
  "votes": 0,
  "comment_count": 0,
  "views": 19,
  "content": "<p>I posted a kernel on how to stream data into your model for those interested.  I approached the problem this way as i found I was having to shrink the pictures down to 64x64 and it seemed I may be losing information (plus who doesn't love large images of cats--see my avatar)</p>\n\n<p>Going to 128x128 did not allow my memory to hold everything so I went and looked for how to do an online-learner neural net.  Low and behold smarter people than me already had this mastered and it is included in keras.</p>\n\n<p><a href=\"https://www.kaggle.com/rdslater/dogs-vs-cats-redux-kernels-edition/data-streaming-for-convoluted-neural-net\">https://www.kaggle.com/rdslater/dogs-vs-cats-redux-kernels-edition/data-streaming-for-convoluted-neural-net</a></p>\n\n<p>I apologize to my friends in the R community--I do this after work so between a job/kids/family I did not have time to find the solution in R.</p>",
  "messages": [
    {
      "id": 159122,
      "postDate": "2017-01-31T23:30:13.447Z",
      "content": "<p>I posted a kernel on how to stream data into your model for those interested.  I approached the problem this way as i found I was having to shrink the pictures down to 64x64 and it seemed I may be losing information (plus who doesn't love large images of cats--see my avatar)</p>\n\n<p>Going to 128x128 did not allow my memory to hold everything so I went and looked for how to do an online-learner neural net.  Low and behold smarter people than me already had this mastered and it is included in keras.</p>\n\n<p><a href=\"https://www.kaggle.com/rdslater/dogs-vs-cats-redux-kernels-edition/data-streaming-for-convoluted-neural-net\">https://www.kaggle.com/rdslater/dogs-vs-cats-redux-kernels-edition/data-streaming-for-convoluted-neural-net</a></p>\n\n<p>I apologize to my friends in the R community--I do this after work so between a job/kids/family I did not have time to find the solution in R.</p>",
      "rawMarkdown": "I posted a kernel on how to stream data into your model for those interested.  I approached the problem this way as i found I was having to shrink the pictures down to 64x64 and it seemed I may be losing information (plus who doesn't love large images of cats--see my avatar)\r\n\r\nGoing to 128x128 did not allow my memory to hold everything so I went and looked for how to do an online-learner neural net.  Low and behold smarter people than me already had this mastered and it is included in keras.\r\n\r\nhttps://www.kaggle.com/rdslater/dogs-vs-cats-redux-kernels-edition/data-streaming-for-convoluted-neural-net\r\n\r\nI apologize to my friends in the R community--I do this after work so between a job/kids/family I did not have time to find the solution in R."
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "159122": "I posted a kernel on how to stream data into your model for those interested.  I approached the problem this way as i found I was having to shrink the pictures down to 64x64 and it seemed I may be losing information (plus who doesn't love large images of cats--see my avatar)\r\n\r\nGoing to 128x128 did not allow my memory to hold everything so I went and looked for how to do an online-learner neural net.  Low and behold smarter people than me already had this mastered and it is included in keras.\r\n\r\nhttps://www.kaggle.com/rdslater/dogs-vs-cats-redux-kernels-edition/data-streaming-for-convoluted-neural-net\r\n\r\nI apologize to my friends in the R community--I do this after work so between a job/kids/family I did not have time to find the solution in R."
  }
}