🚀 go-pugleaf

The following preprint is available from the author: Erwin, H., 1994, 'The
application of Katchalsky network models to radar pattern classification,'
to be published in the Proceedings of the Second Appalachian
Neurodynamics Conference, K. Pribram, ed., INNS Press, LEA, 1994.

Abstract:
  Yong Yao's and Walter Freeman's KII and KIII models of the olfactory
system describe neural network architectures based on the olfactory
system that use chaotic dynamics to perform rapid feature-classification
and novelty-detection. Engineering application of these KI, KII, and
KIII ('Katchalsky') networks has been slowed by a number of issues:
. extension of the published results to practical problems has been
  difficult due to the limited background of most computer engineers
  in the neurophysiology of olfaction,
. training methods for these networks remain poorly understood,
. lack of invariance in biological Katchalsky networks makes data
  encoding for engineering applications unclear.
This paper describes the lessons learned in investigating a KIII
Katchalsky network for an aircraft radar application.

Key Engineering Results:
1. The role of stochastic (chaotic) resonance in supporting
generalization is clarified.
2. The olfactory bulb/anterior olfactory nucleus is identified as a
modified (and robust) Hebb-Marr autoassociator.
3. Synchronization of efferent signals from the olfactory bulb/anterior
olfactory nucleus is identified as the key indicator that this system
has successfully classified an afferent feature complex.
4. The role of the chaos seen in EEG recordings is discussed.

--
Harry Erwin
Internet: herwin@gmu.edu 
Working on Katchalsky nets....