Neural Networks in Recommender Systems with an Optimization to the Neural Attentive Recommender Model

Abstract

The impact of recommender systems on e– commerce, marketing, and user entertainment has long been established. To combat the problem of information overload on the internet, they seek to improve customer-company interactions and provide customers with individualized online product or service recommendations. There are several types of recommender systems, and two of the most common are– Content based & Collaborative filtering-based recommender systems and they are examined in this paper. With the rapid rise in processing efficiency and as the capacity to process deeper neural networks became increasingly feasible, the application of deep learning to recommender systems was inevitable. In this study, we hope to guide the reader through the implementation of Neural networks (a fast deep learning algorithm) to create highly reliable recommender systems using the two aforementioned methodologies. The importance of a "Hybridized" system of recommendation is also illustrated. Finally, we propose a statistical optimization strategy on the Neural Attentive Recommender Model and examine the assumptions, advantages & results in the form of an experimental methodology.