Coders Packet

Sentiment Analysis on IMDB dataset with LSTM on keras

By Vineet Pandey

In this project, we do sentiment analysis on IMDB dataset available on keras using LSTM.

IMDB dataset is readily available on Keras. We are using LSTM model to do sentiment analysis of the movie reviews present in IMDB sentiment analysis. Sentiment analysis means classifying the review as positive or negative. In our model we have used Binary Cross Entropy loss function and Sigmoid Activation function. For optimisation we have used Adam Optimiser. We have truncated/padded all the reviews to max length of 600 words. Also our words dictionary is of top 5000 words. So we have put the index 0 for any word out of those 5000 words.

[For the training, it takes about half an hour on GPU 1050ti].