… Having defined computation as implemented but implementation- independent syntactic
symbol systems, we must ask first whether neural nets fall under this definition. The answer is a …

… to numerous efforts to build machines that simulate the world or communicate with it [Bus70, …
called artificial neural networks. In their most general framework, neural networks consist of …

… of some analog hardware implementations of neural networks. Several criteria for the
classification of general neural networks … The paper also discusses some characteristics of analog …

… analog neural networks called nonlinear resistive networks … world to the analog world,
much more improvement in speed and power efficiency is possible by rethinking neural networks …

… a neural net does not matter if only its computational properties matter. (I think that the proofs
that neural nets … between using a net as hardware to do symbolic computation and training …

… a neural net does not matter if only its computational properties matter. (I think that the proofs
that neural nets … between using a net as hardware to do symbolic computation and training …

… ’t rely on every single synapse or neuron. The power consumption is the major issue in
realizing wafer-scale integration for the presented analog neural network. The number of events …

… world in the same way as biological nervous systems do" [Kohonen-l]. There is a fundamental
difference in the way that conventional digital computers and biological neural networks (…

… of the way in which neural signal processing is performed may guide us to invent scalable
solutions that are able to survive in a presumably ever-shrinking CMOS submicron world. In …

… of analog electronic implementation of artificial neural networks. … analog hardware
implementations are still in the research stage and only a few designs have been applied to “real …