Collaborative Filtering and Deep Neural Networks
Deep Neural Networks provide higher accuracy compared to Shallow Learning on Collaborative Filtering problems. DNN also makes it possible to use additional features to build a recommendation system, providing better recommendations.
Collaborative filtering using embeddings in a Deep Neural Network is the state of the art method to provide product/ service recommendations to users. This is the same technique behind Netflix and Amazon’s recommendations to the users. Additional features from products like specific attributes of the
products can be used to improve accuracy of the recommendation system.
These models can be used in
E-commerce sites to recommend products
Retail outlets with customer loyalty cards to recommend coupons/ offers
Online content providers to recommend appropriate contents to customers