Work-in-Progress: Accelerated Matrix Factorization by Approximate Computing for Recommendation System

Matrix factorization (MF) is widely used in collaborative filtering-based recommendation systems, but the computational complexity greatly increases for larger scaled recommendation systems. We propose to accelerate MF by performing approximate matrix multiplications, considering the joint sparsity of the decomposed matrices. We show our method realizes a more than 1.1 speedup with a minimal error, and the speedup can be higher for the recommendation systems with larger scales.