Genetic Neural Network Trainer

This project shows an example of training a Multilayer Perceptron network with genetic algorithms.

The project's directory structure:

• data/ - various data used for testing
  • data/mnist/ - dataset for handwritten digit recognition
  • data/models/ - models saved after training
• gui/ - graphical user interface implemented using QT
• include/ - main directory with the implementation
  • include/gnnt/mlp/ - compile-time Multilayer Perceptron network
  • include/gnnt/training/ - genetic algorithm implementation
• test/ - test cases for activation functions with the Catch2 framework

Screenshots

The Graphical User Interface

Example prediction for manually drawn digit

Example of training a network