DeepAI AI Chat
Log In Sign Up

Performance of Hyperbolic Geometry Models on Top-N Recommendation Tasks

by   Leyla Mirvakhabova, et al.

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.


page 1

page 2

page 3

page 4


HICF: Hyperbolic Informative Collaborative Filtering

Considering the prevalence of the power-law distribution in user-item ne...

Large-Margin Classification in Hyperbolic Space

Representing data in hyperbolic space can effectively capture latent hie...

HGCC: Enhancing Hyperbolic Graph Convolution Networks on Heterogeneous Collaborative Graph for Recommendation

Due to the naturally power-law distributed nature of user-item interacti...

Scalable Hyperbolic Recommender Systems

We present a large scale hyperbolic recommender system. We discuss why h...

On Automating Triangle Constructions in Absolute and Hyperbolic Geometry

We describe first steps towards a system for automated triangle construc...

Saccadic Eye Movements and the Generalized Pareto Distribution

We describe a statistical analysis of the eye tracker measurements in a ...

Poincaré ResNet

This paper introduces an end-to-end residual network that operates entir...

Code Repositories


Accompanying code for the paper Performance of Hyperbolic Geometry Models on Top-N Recommendation Tasks, accepted at ACM RecSys 2020.

view repo