Path: utzoo!attcan!uunet!snorkelwacker!tut.cis.ohio-state.edu!uc!cs.umn.edu!irani
From: irani@cs.umn.edu (Erach Irani)
Newsgroups: comp.ai.neural-nets
Subject: Neural nets latest trick
Message-ID: <1990Jul18.160459.18705@cs.umn.edu>
Date: 18 Jul 90 16:04:59 GMT
References: <64712@lll-winken.LLNL.GOV> <3071@osc.COM>
Organization: University of Minnesota, Minneapolis - CSCI Dept.
Lines: 25

I agree, neural nets are the latest trick and people are using them
where older techniques do better.

This is good if people do comparisons with the older techniques, but not
many people know about the older techniques INITIALLY (until reviewers,
co-workers ask for a comparison).

And so once they learn about the older techniques they can push the new
techniques (like neural nets) into newer applications (if possible) or
show how they do better/worse on older applications (as compared to older
techniques).

Presumably, it will take time but new applications will be found where
older techniques were not as good once researchers knowing new techniques
learn the older ones once researchers knowing the older techniiques will
bring the researchers with newer techniques into contact with the older
techniques (which I already see happenning with neural nets).

	- erach

-- 
Phone : (Home) (612) 378-2336     Work : (612) 627-4850 
InterNet : irani@cs.umn.edu       UUCP: uunet!umn-cs!irani
Postal Address: Erach A.Irani; 1717, Rollins Ave.; Minneapolis, MN 55414.