Performance of Hyperbolic Geometry Models on Top-N Recommendation Tasks

Abstract

We introduce a simple autoencoder based on hyperbolic geometry for solving standard collaborative filtering problem. In contrast to many modern deep learning techniques, we build our solution using only a single hidden layer. Remarkably, even with such a minimalistic approach, we not only outperform the Euclidean counterpart but also achieve a competitive performance with respect to the current state-of-the-art. We additionally explore the effects of space curvature on the quality of hyperbolic models and propose an efficient data-driven method for estimating its optimal value.

arXiv commentAccepted at ACM RecSys 2020; 7 pages

References in corpus (8)

• Hyperbolic Geometry of Complex Networks
• Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches
• Stochastic gradient descent on Riemannian manifolds
• Bipartite Graphs as Models of Complex Networks
• RecVAE: a New Variational Autoencoder for Top-N Recommendations with Implicit Feedback
• Poincaré Embeddings for Learning Hierarchical Representations
• The Topology of Music Recommendation Networks
• Music Recommendations in Hyperbolic Space: An Application of Empirical Bayes and Hierarchical Poincaré Embeddings

Cited by in corpus (2)

• Personality-Driven Social Multimedia Content Recommendation
• HGCH: A Hyperbolic Graph Convolution Network Model for Heterogeneous Collaborative Graph Recommendation