GLocal-K: Global and Local Kernels for Recommender Systems

TL;DR

GLocal-K introduces a novel kernel-based matrix completion framework for recommender systems, effectively capturing high-dimensional sparse data into low-dimensional features without side information, outperforming existing methods on benchmark datasets.

Contribution

The paper presents GLocal-K, a two-stage kernel-based autoencoder approach that enhances matrix completion in low-resource recommender systems without relying on additional data.

Findings

Outperforms state-of-the-art baselines on ML-100K, ML-1M, and Douban datasets.

Effectively captures item characteristics using global kernels.

Excels in extreme low-resource settings with only rating matrices.

Abstract

Recommender systems typically operate on high-dimensional sparse user-item matrices. Matrix completion is a very challenging task to predict one's interest based on millions of other users having each seen a small subset of thousands of items. We propose a Global-Local Kernel-based matrix completion framework, named GLocal-K, that aims to generalise and represent a high-dimensional sparse user-item matrix entry into a low dimensional space with a small number of important features. Our GLocal-K can be divided into two major stages. First, we pre-train an auto encoder with the local kernelised weight matrix, which transforms the data from one space into the feature space by using a 2d-RBF kernel. Then, the pre-trained auto encoder is fine-tuned with the rating matrix, produced by a convolution-based global kernel, which captures the characteristics of each item. We apply our GLocal-K model under the extreme low-resource setting, which includes only a user-item rating matrix, with no side information. Our model outperforms the state-of-the-art baselines on three collaborative filtering benchmarks: ML-100K, ML-1M, and Douban.