PinaPL

PinaPL Is Not A Projet Long

This is a second implementation of a standard neural network. It uses the forward propagation algorithm for computing outputs, and the backward propagation algorithm for learning.

Classes

The included classes/files are:

• NeuralNetwork, the data structure wrapper with the main algorithms
• Neuron which represents a simple neuron (or perceptron) and transforms data
• NeuralConnection which represents either a connection between two neurons, input and output values, or bias values, and stores data
• Functions which contains the composition, activation and cost functions
• idxParser, the parser for MNIST dataset
• PinaPLDefines and NeuralNetworkDefines which contains some usefull macros
• main, with the testing functions for the XOR operator and the MNIST dataset