https://github.com/alexander-rakhlin/CNN-for-Sentence-Classification-in-Keras

Convolutional Neural Networks for Sentence Classification in Keras
https://github.com/alexander-rakhlin/CNN-for-Sentence-Classification-in-Keras

Last synced: 8 days ago
JSON representation

Convolutional Neural Networks for Sentence Classification in Keras

• Host: GitHub
• URL: https://github.com/alexander-rakhlin/CNN-for-Sentence-Classification-in-Keras
• Owner: alexander-rakhlin
• License: mit
• Created: 2016-03-19T10:49:11.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2018-11-13T05:23:36.000Z (about 6 years ago)
• Last Synced: 2024-10-28T14:34:45.546Z (16 days ago)
• Language: Python
• Size: 6.22 MB
• Stars: 596
• Watchers: 37
• Forks: 205
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Convolutional Neural Networks for Sentence Classification

Train convolutional network for sentiment analysis. Based on "Convolutional Neural Networks for Sentence Classification" by Yoon Kim, [link](http://arxiv.org/pdf/1408.5882v2.pdf). Inspired by Denny Britz article "Implementing a CNN for Text Classification in TensorFlow", [link](http://www.wildml.com/2015/12/implementing-a-cnn-for-text-classification-in-tensorflow/).

For "CNN-rand" and "CNN-non-static" gets to 88-90%, and "CNN-static" - 85%

## Some difference from original article:

* larger IMDB corpus, longer sentences; sentence length is very important, just like data size

* smaller embedding dimension, 20 instead of 300

* 2 filter sizes instead of original 3

* much fewer filters; experiments show that 3-10 is enough; original work uses 100

* random initialization is no worse than word2vec init on IMDB corpus

* sliding Max Pooling instead of original Global Pooling

## Dependencies

* The [Keras](http://keras.io/) Deep Learning library and most recent [Theano](http://deeplearning.net/software/theano/install.html#install) backend should be installed. You can use pip for that. 

Not tested with TensorFlow, but should work.