Autoencoder-Movie-Recommender

We apply autoencoders to the task of making movie recommendations.

1. We began with the Keras example Collaborative Filtering for Movie Recommendations. This example demonstrates Collaborative filtering using the Movielens dataset to recommend movies to users.
2. We construct a feature vector for Tim, who chooses and rates several movies, to generate 10 movie recommendations using the recommender model from the Keras example.
3. To improve our results, we modify the recommender model from the Keras example by replacing the dot product with two hidden Dense layers in our recommender model.
4. We compare the quality of the recommendations from the Keras example and our new model.
5. We improve our recommendations further by applying a Variational Autoencoder to add an additional sampling layer to the network.