SuperSimpleNet - Simple Nerual Network Implementation

SuperSimpleNet is a simple nerual network implementation for educational use and simple classification use cases written in Java. It tries to use parallelization as much as possible to use all available CPU cores for faster processing.

This repository includes use examples in form of Unit tests. There are two examples:

• XOR implementation using neural network
• Classification of samples from ISOLET dataset

Learning curve for ISOLET dataset

Properly classified (mAP): 98.58929143956397 % of training samples
Properly classified (mAP): 95.25336754329699 % of test samples

Getting started

Check unit tests for usage examples. You can run them by running test Gradle task.