pkahn@meridian.ads.com (Phil Kahn) (10/16/87)
My primary field is computer vision. In it, we refer to "Computer vision" which is the theory of vision (independent of implementation). "Machine vision" is vision specifically for applications involving man-made machines. "Natural vision" is reserved for biological vision systems. Applying this to your problem, one may refer to "natural neural nets" or "biological neural nets" (the former can be called "nats" and the latter "bionets"). I rather like the term "bionets". An alternative, which I have long argued for, is to set the terminology right. What most of the literature refers to as neural nets are, in fact, connectionist networks. Neural nets refer to NEURONS, so as you noted, the term should stick. The argument would proceed as thus: neural nets were intended to model biological networks. These original models are not adequate because of [stated reasoning...] Thus, this work proposes a model of neural networks which is more representative of the underlying neurophysiology. This might allow you to share in the current glut in funding for neural nets, and it recognizes the similarity between the correct and bastardized use of the term. One man's opinion...