ImplicitSLIM and How it Improves Embedding-based Collaborative Filtering

TL;DR

ImplicitSLIM introduces an efficient unsupervised method that enhances embedding-based collaborative filtering by leveraging SLIM-like models, leading to better performance and faster convergence without heavy computations.

Contribution

It proposes ImplicitSLIM, a novel approach that extracts embeddings from SLIM-like models efficiently, improving scalability and effectiveness in collaborative filtering.

Findings

Improves performance of collaborative filtering models

Speeds up convergence in training

Reduces memory usage compared to traditional SLIM methods

Abstract

We present ImplicitSLIM, a novel unsupervised learning approach for sparse high-dimensional data, with applications to collaborative filtering. Sparse linear methods (SLIM) and their variations show outstanding performance, but they are memory-intensive and hard to scale. ImplicitSLIM improves embedding-based models by extracting embeddings from SLIM-like models in a computationally cheap and memory-efficient way, without explicit learning of heavy SLIM-like models. We show that ImplicitSLIM improves performance and speeds up convergence for both state of the art and classical collaborative filtering methods. The source code for ImplicitSLIM, related models, and applications is available at https://github.com/ilya-shenbin/ImplicitSLIM.