Basic Reasoning Applicator Implemented as Neural Network (BRAINN)

BRAINN combines rule-based reasoning typical of expert systems, with similarity-based reasoning typical of neural networks, in an attempt to reproduce aspects of human common-sense reasoning. BRAINN's architecture mimics the columnar organisation of the human neocortex. It consists of a large hexagonal network of Hopfield nets. BRAINN encodes knowledge from both rules and a semantic network in a distributed fashion. Each relation in the semantic network is stored in a single Hopfield net, whilst each rule is stored in a set of adjacent Hopfield nets.

The program runs in MS Windows enviroment and has very friendly user interface.

Click here to download executable and source codes of BRAINN

Back to Rafal Bogacz Homepage