/feedforward-neural-network

An implementation of a Multi-Layer Perceptron in C++

Primary LanguageC++MIT LicenseMIT

feedforward-neural-network

A basic implementation of a Multi-Layer Perceptron in C++.

The purpose of this project is to understand the working principle of neural networks by building a network from scratch and feeding it with custom data.

Please check source files for more details:

  • You can find the implementation of various parts of the neural network, going from a simple neuron, to a layer, to the whole network.
  • Both weight and bias neurons are implemented.
  • The topology (depth of the network, along with the number of neurons in each layer) could be specified in the Data.txt file.
  • The data generator actually simulates the behavior of a XOR gate and outputs training data in a non-generic format (Not CSV).

Ressources: