BMF: Block matrix approach to factorization of large scale data

TL;DR

This paper introduces a novel block matrix approach for matrix factorization that efficiently handles large-scale data by operating at the block level, overcoming memory and computational limitations of traditional methods.

Contribution

The paper proposes a new block matrix factorization technique that improves scalability and efficiency for large-scale data on CPU and GPU architectures.

Findings

Enables factorization of matrices larger than available memory

Reduces computational time compared to traditional methods

Applicable to both CPU and GPU environments

Abstract

Matrix Factorization (MF) on large scale matrices is computationally as well as memory intensive task. Alternative convergence techniques are needed when the size of the input matrix is higher than the available memory on a Central Processing Unit (CPU) and Graphical Processing Unit (GPU). While alternating least squares (ALS) convergence on CPU could take forever, loading all the required matrices on to GPU memory may not be possible when the dimensions are significantly higher. Hence we introduce a novel technique that is based on considering the entire data into a block matrix and relies on factorization at a block level.