Speeding-Up Back-Propagation in DNN: Approximate Outer Product with Memory
This work addresses the bottleneck of training time for deep learning practitioners, though it appears incremental as it builds on existing gradient descent methods with a novel approximation technique.
The paper tackles the computational complexity of back-propagation in deep neural network training by proposing the Mem-AOP-GD algorithm, which approximates gradient descent using a subset of outer products and corrects bias with memory accumulation, resulting in significant improvements in both computational efficiency and accuracy.
In this paper, an algorithm for approximate evaluation of back-propagation in DNN training is considered, which we term Approximate Outer Product Gradient Descent with Memory (Mem-AOP-GD). The Mem-AOP-GD algorithm implements an approximation of the stochastic gradient descent by considering only a subset of the outer products involved in the matrix multiplications that encompass backpropagation. In order to correct for the inherent bias in this approximation, the algorithm retains in memory an accumulation of the outer products that are not used in the approximation. We investigate the performance of the proposed algorithm in terms of DNN training loss under two design parameters: (i) the number of outer products used for the approximation, and (ii) the policy used to select such outer products. We experimentally show that significant improvements in computational complexity as well as accuracy can indeed be obtained through Mem-AOPGD.