This repository contains the server code for my Deep Learning Recommender Systems talk. As a disclaimer, it is likely not set up well to run without changes being made, and is not incredibly well-designed. That said, it works, and the code may still be helpful in understanding how the models were deployed behind a web server.
The project is based on data from MovieLens, code from Robin Devooght, and these two papers:
Devooght, Robin and Hugues Bersini "Collaborative Filtering with Recurrent Neural Networks" (2017)
https://arxiv.org/abs/1608.07400
Kuchaiev, Oleksii and Boris Ginsburg "Training Deep Autoencoders for Collaborative Filtering" (2017)
https://arxiv.org/abs/1708.01715
The code for implementing the collaborative filtering autoencoder in Keras is in the IPython notebook.
To run the server:
$ FLASK_APP=app.py flask run
✨🍰✨