GLocal-K: Global and Local Kernels for Recommender Systems
This addresses the problem of improving recommendation accuracy in low-resource settings without side information for users and platforms, though it appears incremental as it builds on kernel and autoencoder methods.
The paper tackles the challenge of matrix completion for recommender systems using high-dimensional sparse user-item matrices by proposing GLocal-K, a global-local kernel-based framework that transforms data into a low-dimensional feature space, and it outperforms state-of-the-art baselines on benchmarks like ML-100K, ML-1M, and Douban.
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.