Neural Network

This project implements a simple neural network with the following features:

• learning via backpropagation
• cost functions: quadratic / cross-entropy
• activation function: sigmoid / tanh

The project also contain examples of experiments that use the neural network: please refer to source code in com.penna.neural.experiments for more information.

Compiling

Install Maven and then execute on the project root

$ mvn clean package

Running sample experiment

From project root:

$ cd src/main/scripts
$ sh <experiment runner>

where <experiment runner> is a shell script that executes one of the experiments in the script directory.

Further information and questions

email to [email protected]