Using Deep Learning Neural Networks to classify reviews of movie dataset to Positive and Negative Sentiment.
We are using keras that act as a Wrapper on top of Theano/Tensorflow to create ML models easily as creating models using Theano or Tensorflow requires a lot of code to be written.
Requirements -
- Python 3
- Google word vectors (https://code.google.com/archive/p/word2vec/)
- Theano/Tensorflow (I have created model using Theano)
- Keras (As a wrapper around Theano/Tensoflow)
Here we have used LSTM that are best RNN for doing text classification. Its a binary class problem i.e positive and Negative sentiment. I was able to get 90% accuracy. But we can improve it more my creating more complex model and tuning the hyper parameters.
Just run Keras-LSTM-Sentiment-classification.ipynb notebook and check the results. Happy Learning.