POGD: Gradient Descent with New Stochastic Rules

TL;DR

POGD is a novel optimization algorithm combining gradient descent with particle swarm principles, demonstrating adaptive learning, faster training, and better local minimum avoidance in CNN image classification tasks.

Contribution

This paper introduces POGD, a new stochastic optimization method that integrates PSO principles into gradient descent, enhancing training speed and robustness.

Findings

Faster convergence to target accuracy in CNN training

Improved ability to avoid local minima

Adaptive learning capability demonstrated on MNIST and CIFAR-10

Abstract

There introduce Particle Optimized Gradient Descent (POGD), an algorithm based on the gradient descent but integrates the particle swarm optimization (PSO) principle to achieve the iteration. From the experiments, this algorithm has adaptive learning ability. The experiments in this paper mainly focus on the training speed to reach the target value and the ability to prevent the local minimum. The experiments in this paper are achieved by the convolutional neural network (CNN) image classification on the MNIST and cifar-10 datasets.