Multi Layer Perceptron

Implementation of a basic multilayer perceptron using C++ without any external libraries.
framework is used for GUI, project is written according to MVC design pattern.

Features 🤯

• Learning on given training dataset using cross-validation
• Testing trained model with built-in drawing field or by importing pre-drawn image
• Configure model properties for training such as:
  • Perceptron implementation type
    • Matrix
    • Graph
  • Number of hidden layers
  • Number of neurons (units) per hidden layer
  • Learning rate
  • Activation Function
    • Sigmoid
    • Leaky ReLU
    • Linear
    • Bipolar Sigmoid
  
• Import pretrained model
• Testing trained model on testing dataset

Installation ⚙️

   
   make install
   

And that's it! Compiled application will be located in the bin directory.

On MacOS it will also be located in Launchpad for ease of use.

On Windows... Well, we are working on it!

Requirements ☑️

• C++17
• Qt6.5.0

TODO 👨‍💻

• Support any dataset by using mapping
• Change pen's color for different datasets
• Create both portable and installable applications for Windows

Acknowledgements 🙇

• Dmitriy Korobchenko for video explaining maths behind back propagation algorithm