Fast Non-Bayesian Poisson Factorization for Implicit-Feedback Recommendations

11/05/2018
by   David Cortes, et al.
0

This work explores non-negative matrix factorization based on regularized Poisson models for recommender systems with implicit-feedback data. The properties of Poisson likelihood allow a shortcut for very fast computation and optimization over elements with zero-value when the latent-factor matrices are non-negative, making it a more suitable approach than squared loss for very sparse inputs such as implicit-feedback data. A simple and embarrassingly parallel optimization approach based on proximal gradients is presented, which in large datasets converges 2-3 orders of magnitude faster than its Bayesian counterpart (Hierarchical Poisson Factorization) fit through variational inference techniques, and 1 order of magnitude faster than implicit-ALS fit with the Conjugate Gradient method.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
06/01/2020

Ordinal Non-negative Matrix Factorization for Recommendation

We introduce a new non-negative matrix factorization (NMF) method for or...
research
04/13/2016

Hierarchical Compound Poisson Factorization

Non-negative matrix factorization models based on a hierarchical Gamma-P...
research
11/07/2013

Scalable Recommendation with Poisson Factorization

We develop a Bayesian Poisson matrix factorization model for forming rec...
research
10/26/2021

iALS++: Speeding up Matrix Factorization with Subspace Optimization

iALS is a popular algorithm for learning matrix factorization models fro...
research
12/18/2015

Deep Poisson Factorization Machines: factor analysis for mapping behaviors in journalist ecosystem

Newsroom in online ecosystem is difficult to untangle. With prevalence o...
research
06/16/2017

Adversarial Variational Inference for Tweedie Compound Poisson Models

Tweedie Compound Poisson models are heavily used for modelling non-negat...

Please sign up or login with your details

Forgot password? Click here to reset