Learning by Glueing

This site is about some algorithms, programs and ideas. Some time ago I stumbled upon the book ‘On Intelligence’ by Jeff Hawkins, on a bookshelf in my local library. I, too, was disappointed with the state of artificial intelligence and the many approaches which again and again have led from initial euphoria into dead ends.

Suddenly I realized that my work might be new,
that there is no scholarly consensus on how the brain works,
and that my simple programs and ideas might be the missing links.

Since then, I’ve built a simple OCR Application and a simple audio processor.
There is also a Pdf ( in german, sorry ) that contains more or less the same material.
Some source code. Requirements: QT4, g++ .
some image files ( 140 Mbyte )

New source code.
This version uses 64 engines with |Y|=2 in parallel. The weight of a vote for ’0′ is simply the low pass filtered outcome. The average activation is regulated towards 0.5 and information transmission is maximised. This is also physiologically plausible. The entire engine consists solely of digital memory, comparators and integrators. Even the light sensors fit into this framework.

Read more at: Classifiers   or   Discrete Categories and Continuous Change

Post a Comment