FYI -- this is moderately interesting
Not an AGI approach, but potentially a tool that could be used within one...
-- Ben
Hi Ben,
I presume you've come across the "AND Corporation" http://www.andcorporation.com/frame_technology.html
They claim to have a technology that is vastly better than traditional neural
networks:
" The HNeT system allows our developers to construct neuron cell
assemblies, and integrate these neural assemblies into applications. The core
of the HNeT system is a Dynamic Link Library (DLL) containing over 90
functions for creation of cell assemblies, and customization of cells.
Employing holographic principles, HNeT cells provide both real-time learning
and dramatic improvements in performance over structurally more complex
back-propagation / genetic neural networks. Holographic / quantum neural
technology provides an exceptionally high "connection per second" or CPS
rating; in excess of 40 Million CPS on Pentium III processors. This allows an
HNeT cell assembly to learn and respond to several thousand input patterns
in under a second."
Comparative performance is discussed at: http://www.andcorporation.com/technology_3a.html and the science/method behinnd the system is discussed at: http://www.andcorporation.com/frame_math.html http://www.andcorporation.com/frame_biology.html
Is any of this relevant to the Novamente project?
Cheers, Philip
-------
To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
