DemoNN

A neural network for spiral classification in python using numpy Includes:

• L1 & L2 regularization
• Different activation functions
• SGD with momentum
• Mini batch implementation

Main.py

• Read coirdinates input files from CSV, build neural network with constant configuration.
• can read a serialized network class from saved file

Netowrk.py

A neural network class Perform train and test functionality.

Layer.py

Includes:

• FeedForward algorithm
• BackPropogation algorithm
• Calculation of delta vector for weights change in backprop
• Bias term

activationfunction.py

Enable different activation functions for the network

Incstructions

• If you want to train the netwrok call testSpiralClassification
• If you want to test serialized network call RunSerializedNetwork