bernlewal / rapid-j

Rapid Algebrahic self-Programming with Incremental Deductions / implemented in Java

Rapid/J

A machine-learning network based on self-constructing graphs of neurons during the supervised learning process. The interconnected neurons implement algebrahic-functions for evaluating the weights and firing functions.

(Rapid/J = Rapid Algebrahic self-Programming with Incremental Deductions / implemented in Java)

Goals / specific features:

• network/graph is built-up during learning
• needs only mimimum of learning-data
• network/graph is self optimizing
• neuronal-network function is transparent; GraphML export
• (neurons are working independently (the final goal) )
• (data/information is passing "as waves" through the network/graph)

Attention: work in progress - project is in an early stage.

Getting started:

1. Clone the whole repository
2. Open the Project in NetBeans
3. Run SkalarTestSuite.java in test/rapid/net/skalar

HOWTO: program your own neural-network

1. open test/rapid/net/Numbers123Test.java
2. follow the instructions inside the source-file