{
  "id": 93552,
  "title": "Local Minimum",
  "url": "/competitions/histopathologic-cancer-detection/discussion/93552",
  "author_name": "sumanthmeenan",
  "post_date": "2019-05-28T05:50:00.551000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>while training my model for 20 epochs using adam optimiser, binary cross entropy loss function and accuracy metric, my model is strucking at local minimum as accuracy is no more increasing than  59% may be because i used Dropout = 0.2 for 3 convolutional layers and 3 hidden layers,  . \nSo, later on i tried removing drop out layers, and after 70 epochs my models training accuracy is 82%, while validation accuracy is 80%. </p>",
  "messages": [
    {
      "id": 538088,
      "postDate": "2019-05-28T05:50:00.550Z",
      "content": "<p>while training my model for 20 epochs using adam optimiser, binary cross entropy loss function and accuracy metric, my model is strucking at local minimum as accuracy is no more increasing than  59% may be because i used Dropout = 0.2 for 3 convolutional layers and 3 hidden layers,  . \nSo, later on i tried removing drop out layers, and after 70 epochs my models training accuracy is 82%, while validation accuracy is 80%. </p>",
      "rawMarkdown": "while training my model for 20 epochs using adam optimiser, binary cross entropy loss function and accuracy metric, my model is strucking at local minimum as accuracy is no more increasing than  59% may be because i used Dropout = 0.2 for 3 convolutional layers and 3 hidden layers,  . \nSo, later on i tried removing drop out layers, and after 70 epochs my models training accuracy is 82%, while validation accuracy is 80%. ",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "538088": "while training my model for 20 epochs using adam optimiser, binary cross entropy loss function and accuracy metric, my model is strucking at local minimum as accuracy is no more increasing than  59% may be because i used Dropout = 0.2 for 3 convolutional layers and 3 hidden layers,  . \nSo, later on i tried removing drop out layers, and after 70 epochs my models training accuracy is 82%, while validation accuracy is 80%. "
  }
}