Accelerating block coordinate descent for nonnegative tensor factorization

This paper is concerned with improving the empirical convergence speed of block-coordinate descent algorithms for approximate nonnegative tensor factorization (NTF). We propose an extrapolation strategy in-between block updates, referred to as heuristic restarts (HER). HER significantly accelerates most existing dense NTF, in particular challenging computational scenarios, while requiring a negligible additional budget.