Neural Collaborative Filtering vs. Matrix Factorization Revisited

TL;DR

This paper critically reevaluates neural collaborative filtering, demonstrating that simple dot products outperform learned MLP-based similarities in recommendation tasks, and discusses practical implications for deployment.

Contribution

It shows that with proper tuning, dot products outperform MLPs in collaborative filtering, and highlights the practical and theoretical challenges of using MLPs as embedding comparators.

Findings

Proper hyperparameter tuning favors dot products over MLPs.

Learning dot products with MLPs is non-trivial despite their universal approximation capability.

Dot products enable more efficient retrieval algorithms suitable for production environments.

Abstract

Embedding based models have been the state of the art in collaborative filtering for over a decade. Traditionally, the dot product or higher order equivalents have been used to combine two or more embeddings, e.g., most notably in matrix factorization. In recent years, it was suggested to replace the dot product with a learned similarity e.g. using a multilayer perceptron (MLP). This approach is often referred to as neural collaborative filtering (NCF). In this work, we revisit the experiments of the NCF paper that popularized learned similarities using MLPs. First, we show that with a proper hyperparameter selection, a simple dot product substantially outperforms the proposed learned similarities. Second, while a MLP can in theory approximate any function, we show that it is non-trivial to learn a dot product with an MLP. Finally, we discuss practical issues that arise when applying MLP based similarities and show that MLPs are too costly to use for item recommendation in production environments while dot products allow to apply very efficient retrieval algorithms. We conclude that MLPs should be used with care as embedding combiner and that dot products might be a better default choice.