Analog VLSI and Neural Systems (Hardcover)

~ (Author)

Editorial Reviews

Product Description

The first book to take VLSI into the analog domain and apply it to biology. It provides solid tools for research in artificial intelligence and neurobiology while illustrating powerful new applications for analog systems.

From the Back Cover

This book introduces Mead's pioneering work on the design of neural networks and their implementation in analog VLSI systems. Mead observes first that the nervous systems of even simple animals contain computing paradigms far more effective than any found in currently available computers. He then shows how the powerful organizing principles of these nervous systems can be realized in silicon integrated circuits. His examples include silicon neural systems that replicate animal senses - chips that can see and hear!

Product Details

• Hardcover: 371 pages
• Publisher: Addison Wesley Publishing Company; 1st edition (January 1, 1989)
• Language: English
• ISBN-10: 0201059924
• ISBN-13: 978-0201059922
• Product Dimensions: 9.1 x 6.2 x 0.8 inches
• Shipping Weight: 1.4 pounds
• Average Customer Review: 4.5 out of 5 stars  See all reviews (4 customer reviews)
• Amazon.com Sales Rank: #752,065 in Books (See Bestsellers in Books)

  Popular in these categories: (What's this?)

  #12 in	Books > Professional & Technical > Engineering > Electrical & Electronics > Very-Large-Scale Integration (VLSI)
  #26 in	Books > Computers & Internet > Computer Science > Circuitry > Analog Electronics

Customer Reviews

Most Helpful Customer Reviews

8 of 8 people found the following review helpful:

4.0 out of 5 stars The classic text for this kind of circuitry., June 7, 1997

By A Customer

Assuming you know something about CMOS and VLSI design, this is the classic text to cover the broad base to get started in understanding how one goes about designing actual hardware for various neural network architectures. Both analog and digital approaches are discussed, and the circuits are clearly explained with lots of schematics and plenty of derivative mathematics that show why a particular approach has utility for a given problem. There are a lot of new books (Mead has a new one out) but they owe a large debt to this book.

1 of 1 people found the following review helpful:

5.0 out of 5 stars make silicon eyes and ears, March 7, 2007

By	W Boudville (Terra, Sol 3) - See all my reviews (TOP 10 REVIEWER)    (REAL NAME)

The book describes an interesting niche in VLSI design. Most VLSI chips implement digital logic. But Mead took a different tack, emphasising the analog mode of operation of the transistors. In most digital electronics texts, this regime of current-voltage performance is mostly cursorily dealt with.

What Mead did was use this often where the current through the source and drain was some exponential function of the voltage at the transistor gate. An oversimplification, perhaps, but it captures the essence of the book. By tying together transistors, Mead was able to build circuits that emulated the performance of the eye and ear. The text then uses these to make silicon chips that might mimic the biological sensors.

The book also embodies Mead's approach to understanding the brain and its neural networks. He claims that the problem is very hard. And that we can usefully make progress by looking at the brain's input sensors. As these are much simpler to understand and implement.

Mead carried the ideas here into Synaptics. A Silicon Valley startup that he co-founded.

Sadly, the book is out of print. (Why??) The prices of $129 and higher by third party sellers are way excessive.

4.0 out of 5 stars absolute classic, though kind of dated, October 19, 2009

By	Jonathan - See all my reviews

Amazon Verified Purchase

Carver Mead was one of the of the pioneers of the idea of a "silicon neuron" back in the 1980's. The interesting part is that he sees the direct analog between synthetic neurons and conductance-based models. Instead of membrane capacitance being charged by ions in the neurons, VLSI capacitors are charged by electrons, and so forth.

By now, the book is somewhat dated - a lot of advances have happened since it was published - but it's still a classic, and highly recommended for anyone interested in the possibility of using analog computing for modeling neurons and networks.