MLP

A simple implementation of Multi Layer Perceptron(MLP) based on Object Oriented Programming with python.

Using backpropagation with gradient descent to update weights.

Hyper-parameters can be set manually:

• Number of hidden layers
• Number of perceptrons for each hidden layer
• Active function for each hidden layer
• Epoch No.
• Learning rate (η)

Files

MLP.py: Implementation of mlp
main.py: An exmaple usage of code on iris dataset

TODO:

• move active functions to mlp.py structure
• add bias
• implementing momentum