TatsukiSerizawa / LDWPSO-CNN

Paper "Optimization of Convolutional Neural Network Using the Linearly Decreasing Weight Particle Swarm Optimization" Program

LDWPSO CNN

Python implementation of LDWPSO CNN (Linearly Decreasing Particle Swarm Optimization Convolutional Neural Network).
The program rewrites and uses part of the Hyperactive library.

Article: Optimization of Convolutional Neural Network Using the Linearly Decreasing Weight Particle Swarm Optimization

Usage

Requirement

• Python 3.8
• scikit leran 1.0
• keras 2.8

LDWPSO CNN

1. Clone this repository

2. The baseline LeNet-5 can be run with the following command

   python src/lenet5.py
   

3. If you use MNIST dataset, you can use the following command.

   python src/pso_mnist.py
   

4. If you use CIFAR-10 dataset, you can use the following command.

   python src/pso_cifar10.py
   

Validation

The output parameters can be verified by rewriting the following program.

• src/pso_mnist_result.py
• src/pso_cifar10_result.py

Use data

• MNIST
• CIFAR-10