Scaling transition from momentum stochastic gradient descent to plain stochastic gradient descent

TL;DR

This paper introduces TSGD, a new optimization method that transitions from momentum SGD to plain SGD, combining fast training and high accuracy with a decreasing learning rate for stable convergence.

Contribution

The paper proposes a novel scaling transition from momentum SGD to plain SGD with a linearly decreasing learning rate, enhancing training speed and accuracy.

Findings

Faster training speed compared to traditional SGD and momentum SGD.

Higher accuracy achieved with the TSGD method.

Improved stability during training.

Abstract

The plain stochastic gradient descent and momentum stochastic gradient descent have extremely wide applications in deep learning due to their simple settings and low computational complexity. The momentum stochastic gradient descent uses the accumulated gradient as the updated direction of the current parameters, which has a faster training speed. Because the direction of the plain stochastic gradient descent has not been corrected by the accumulated gradient. For the parameters that currently need to be updated, it is the optimal direction, and its update is more accurate. We combine the advantages of the momentum stochastic gradient descent with fast training speed and the plain stochastic gradient descent with high accuracy, and propose a scaling transition from momentum stochastic gradient descent to plain stochastic gradient descent(TSGD) method. At the same time, a learning rate that decreases linearly with the iterations is used instead of a constant learning rate. The TSGD algorithm has a larger step size in the early stage to speed up the training, and training with a smaller step size in the later stage can steadily converge. Our experimental results show that the TSGD algorithm has faster training speed, higher accuracy and better stability. Our implementation is available at: https://github.com/kunzeng/TSGD.