Key points are not available for this paper at this time.
At present, computers cannot think, make independent decisions, recognize visual and audible patterns, reach reliable conclusions without complete information or learn from their mistakes. The question this book and research into artificial neural networks poses is: what if they could? Exploring the possibility of computers thinking, this work looks at how experimental computers are being designed today to meet tomorrow's anticipated demand for unsupervised high-tech automation. Under the neural model, computers will perform brain-style processing through a network of neurons, each carrying out separate and distinct operations simultaneously. This book explains how this process works, how neural networking technology has evolved, how it differs from conventional CPU-based computing and how hardware and software developers are implenting it. Topics covered include: the McCulloch-Pitts artificial neuron - the birth of networking; Hopfield and Boltzmann networks - early progress in neural research; can computers learn? - the principles of error propagation; new computer hardware for implementing neural networking theories and neural networking and cognitive science - how can connectionist computing help us better understand the human brain? The accompanying disk includes programs for simulating artificial neural networks.
A Thu, study studied this question.