On Accelerating the Regularized Alternating Least Square Algorithm for Tensors
For researchers working on tensor decomposition, this work offers an incremental improvement in convergence speed for the RALS algorithm.
The paper proposes a fast iterative method using Aitken-Stefensen-like updates to accelerate the regularized alternating least square (RALS) algorithm for tensor approximation, achieving faster convergence than standard and regularized ALS in numerical experiments.
In this paper, we discuss the acceleration of the regularized alternating least square (RALS) algorithm for tensor approximation. We propose a fast iterative method using a Aitken-Stefensen like updates for the regularized algorithm. Through numerical experiments, the fast algorithm demonstrate a faster convergence rate for the accelerated version in comparison to both the standard and regularized alternating least squares algorithms. In addition, we analyze the global convergence based on the Kurdyka- Lojasiewicz inequality as well as show that the RALS algorithm has a linear local convergence rate.