Neural Networks Books

This book contains interesting LISP source code. The title say about Neural Networks and chaos, but the Chaos chapter is very small. However, this book contains a game!!. The Chess program (not use neural networks) is written totally in LISP. And play good. You can study the search techniques. I recommend this book for LISP lovers.

Excellent book. I highly recommend it. Although Computational Intelligence could include almost any subject, this book is a comprehensive review of the most agreed-upon paradigms in CI. Haven't checked out the CD that comes with the book, so I cannot comment about it.

Chapter 4 of this book is the pot of gold: A concrete, detailed description of how the cerebral cortex works. Thinking relies upon an operation which could be called confabulation (in the chapter it is called concensus building). THIS IS NOT REASONING (at least in any classical sense). Yet, because the simple kind of knowledge used (antecedent support probabilities) is exhaustive, concensus building yields excellent conclusions. This cortical theory also shows why AI has failed: reasoning is too difficult and requires too much knowledge of an expensive type. Cortex gets by with a much simpler type of knowledge (which only concerns pairs of object and action attributes, not n-tuples) which, while it is needed in huge quantities, is easy to obtain. An implication of this corticl theory is that we can now proceed to develop successful AI by adopting this cortical design. The theory is illustrated by means of computer thinking experiments that yield compelling results (and which readers can replicate).

Understanding the brain is a research project with a very long history - as long as the history of civilization. Despite thousands of years of effort, it has only been within the last two decades that "understanding the brain" has more than a mythical or philosophical meaning.

The reason for this is the computer. Just as physics changed from an essentially mediaeval natural philosophy to a modern science through photography, so neuroscience has only come into being through the use of computers. I am not suggesting this in a metaphorical sense. Human brains do not function like computers. But computers have provided a way of modeling processes of nervous systems with increasing verisimilitude.

Consider a phenomenon that exists within at least 12 orders of magnitude; that has an evolutionary history of several billion years; that embraces information from elementary particle physics to cell biology to physiology to psychology to sociology to cosmology (and I am leaving out many other, no less instrumental studies, e.g. linguistics, literature, art) . Consider that no model of this phenomenon has ever survived the age in which it was devised. Consider that even now we do not have an agreed upon terminology for describing its physical characteristics at a gross anatomical level. These are some of the most obvious hurdles that one need overcome if one is to begin "understanding the brain".

Given the astonishing degree of complexity that is the human brain, what is it that is possible with computers that has not been possible before? Computers, and specifically computers used in neuroinformatics, allow us to store, organize and retrieve information. They allow us to build dynamic models, and to test these models with simulated experiments. They allow us, also for the first time in history, to image, in a noninvasive, physiologically tender manner, the workings of living brains. They allow us to talk to one another around the world at any time, in whatever mode of communication is most convenient or salient. But perhaps most importantly, computers provide a tool for grappling with nonlinear causality.

When chaos was first observed in a rigorous fashion, it was thought to be an exotic function of complex systems. But take a closer look. Chaos - and nonlinearity- are now known to be fundamental facts of Nature. Nature is more creative than we could imagine.

Arbib and Grethe have mapped out a research strategy which is one of the first coherent such strategies in neuroscience. They have taken on the orders of magnitude problem, the multi-discipline problem, the modeling problem, etc. and have provided a trajectory through these problems which permits an organized body of knowledge to be built. For that reason, their book is foundational and generative of neuroscience in a legitimately scientific way. If a theory of the brain is possible, then it will come about somewhat in the manner they have laid out. They have made explicit what has been occult for twenty years.

For any student with a serious interest in learning about the brain, this is the book to start with, whether that student is an enthusiastic amateur or a seasoned researcher.

This book has been a goldmine of ideas in control theory and neural networks for my research and has contributed greatly to my knowledge of neural networks. I can say that about very few other books. The papers on system identification using neural networks have been particularly useful to me.

just got it, flipped it open to a random page, started reading, learned something. It's very well written, explaining things very clearly, with figures to aid in explanations. seems like an excellent book so far.

"Diversity in the Neuronal Machine" provides valuable insights into the topic of interneurons (formally classified by the author as "using GABA as its primary fast neurotransmitter in the normal, non-pathological state") as opposed to the principal excitatory projection neurons of the nervous system. Beginning with the studies and hypotheses of the Nobel laureate Santiago Ramon y Cajal, Dr Soltesz explores the roles of "the so-called neurons with short axons..." as the "elemental components" of cerebral activity.

Interneuronal complexity is clearly explained by an in-depth evaluation of several classification schemes that demonstrate the existence of a large variety of interneuronal types. This leads the author to propose a statistical quantification of interneuronal diversity. The question of whether this diversity could explain the evolution of higher intelligence is addressed by the novel applications of network theory to neuronal microcircuit function. Using both mammalian and invertebrate models and computer simulations as examples, Cajal's hypotheses on functional neuroanatomy are extended to incorporate cell-to-cell interneuronal variability as an important factor in the study of interneuronal heterogeneity.

In all, this book provides a compelling insight into the relatively unexplored subject of interneurons and their physiological and even evolutionary significance. An added bonus is the inter-disciplinary nature of this book which opens the neuroscientist to novel insights emanating from the fields of computer modeling, network theory and even artificial intelligence. Fascinating reading!!!

This book provides a nice balance of math, examples, and MATLAB computer. It was of great help for me to understand and code up my own forms of ANNs. The example code provided is a really nice feature. It also puts a great deal of emphasis on relating statistics with ANNs.

This title is a project review of a part of world-wide project named "Real World Computing (RWC) Program" organized by Ministry of Economy, Trade and Industry of Japan in 1990's. The book focuses on the theoretical issue, which covers genetic algorithm, Bayesian inference, mean field approximation, computational learning theory, mobile robots and so on. The quality is very high because the contents are based on high quality journal papers and conference papers. It is also very understandable for non-specialist readers.

As someone who is interested in the subject of computational neuroscience, I found this book of great help. It is the best introduction to this very interesting and new discipline and well prepares the reader for more advanced reading of the subject.