AUC Optimisation and Collaborative Filtering

TL;DR

This paper introduces a differentiable class of objective functions for optimizing AUC in recommendation systems, focusing on top-ranked items, with scalable algorithms and theoretical analysis of consistency and generalisation.

Contribution

It proposes a novel smooth surrogate for AUC tailored for matrix factorization, with a scalable stochastic gradient method and theoretical insights into its consistency and generalisation.

Findings

The proposed method effectively optimizes AUC in recommendation tasks.

Scalable algorithms perform well on synthetic and real datasets.

Theoretical analysis supports the method's consistency and generalisation capabilities.

Abstract

In recommendation systems, one is interested in the ranking of the predicted items as opposed to other losses such as the mean squared error. Although a variety of ways to evaluate rankings exist in the literature, here we focus on the Area Under the ROC Curve (AUC) as it widely used and has a strong theoretical underpinning. In practical recommendation, only items at the top of the ranked list are presented to the users. With this in mind, we propose a class of objective functions over matrix factorisations which primarily represent a smooth surrogate for the real AUC, and in a special case we show how to prioritise the top of the list. The objectives are differentiable and optimised through a carefully designed stochastic gradient-descent-based algorithm which scales linearly with the size of the data. In the special case of square loss we show how to improve computational complexity by leveraging previously computed measures. To understand theoretically the underlying matrix factorisation approaches we study both the consistency of the loss functions with respect to AUC, and generalisation using Rademacher theory. The resulting generalisation analysis gives strong motivation for the optimisation under study. Finally, we provide computation results as to the efficacy of the proposed method using synthetic and real data.