Paper on instantaneously trained neural networks

[Subhash Kak]

The following paper on instantaneous learning and its
applications to time-series prediction and metasearch
engine design is available at:

http://www.ee.lsu.edu/kak/x5kak.lo.pdf

------------------------------

Subhash Kak, Faster web search and prediction using instantaneously 
trained neural networks, IEEE Intelligent Systems, 
vol. 14, pp. 79-82, November/December 1999.

Abstract:

Over the past few years, we have developed new neural 
network designs that model working memory in their ability
to learn and generalize instantaneously.  These networks 
are almost as good as backpropagation in the quality of 
their generalization.  With their speed advantage, they can 
be used in many real-time applications of signal processing, 
data compression, forecasting, and pattern recognition.
In this paper, we describe the networks and their applications to
two problems: (1) prediction of time-series; (2) design of an 
intelligent metasearch engine for the Web. The description of
these two applications will provide enough information to see 
how they could be used in other situations.