» » Neural Networks and Intellect: Using Model-Based Concepts
Neural Networks and Intellect: Using Model-Based Concepts

Neural Networks and Intellect: Using Model-Based Concepts by Leonid I. Perlovsky

ISBN10: 0195111621
ISBN13: 978-0195111620
Author: Leonid I. Perlovsky
Book title: Neural Networks and Intellect: Using Model-Based Concepts
Publisher: Oxford University Press; 1 edition (October 19, 2000)
Language: English
Category: Computer Science
Size PDF: 1116 kb
Size ePub: 1691 kb
Size Fb2: 1431 kb
Rating: 4.3/5
Votes: 831
Pages: 496 pages

Neural Networks and Intellect: Using Model-Based Concepts by Leonid I. Perlovsky

Neural Networks and Intellect: Using Model-Based Concepts describes a new mathematical concept of modeling field theory and its applications to a variety of problems. Examining the relationships among mathematics, computations in neural networks, signs and symbols in semiotics, and ideas of mind in psychology and philosophy, this unique text discusses deep philosophical questions in detail and relates them to mathematics and the engineering of intelligence. Ideal for courses in neural networks, modern pattern recognition, and mathematical concepts of intelligence, it will also be of interest to anyone working in a variety of fields including neural networks, AI, cognitive science, fuzzy systems, pattern recognition and machine/computer vision, data mining, robotics, target tracking, and financial forecasting. Neural Networks and Intellect describes model-based neural networks that utilize the intriguing concept of an internal "world" model, an idea that originated in artificial intelligence and cognitive psychology but whose roots date back to Plato and Aristotle. Combining the a priori knowledge with adaptive learning, the new mathematical concept addresses the most perplexing problems in the field of neural networks: fast learning and robust generalization. The author provides an overview of computational intelligence and neural networks, relating hundreds of seemingly disparate techniques to several fundamental mathematical concepts, which are in turn linked to concepts of mind in philosophy, psychology, and linguistics. Topics covered include the hierarchical and heterarchical organization of intelligent systems, statistical learning theory, genetic algorithms, complex adaptive systems, mathematical semiotics, the dynamic nature of symbols, Godel theorems and intelligence, emotions and thinking, the mathematics of emotional intellect, consciousness, and more. Perlovsky's remarkable conclusion is that the work of ancient philosophers came closer to the computational concepts emerging today than that of pattern recognition and AI experts of just a few years ago. The following website contains information about Dr. Perlovsky's current research related to the theory developed in the book and about available funding opportunities under a Research Associateship Program: to find it search for Perlovsky on http://www4/nationalacademies.org/pga/rap.nsf. Other sources of funding might be available for US-based and international researchers.


This is an important book, but it would be more useful if it didn't have so many errors in the equations. Chapter 5, for example, states that the rates are a part of the model, but the equations on pages 210-213 only define the rates, but never use them. Equation 5.2-11 should have det(Cbar) in the numerator. There are many more errors like these in just the MLANS description in Chapter 5.

Artificial intelligence research goes back to the 1940s as do the first developments in electronic computers. Perlovsky traces the history of AI in a thoughtful and scholarly manner, emphasizing his philosophy and his own generalization of the theory which he calls Modeling Field Theory (MFT). He also traces the study of intelligence back to the Greek philosophers beginning with Plato and Aristotle some 2300 years ago.
The book however, provides more than just a philosophy for artificial intelligence. It mixes in some very important mathematics from the disciplines of engineering, statistics and computer science. Mathematical techniques and models have been particularly useful in the solution to problems in classification, clustering, pattern recognition, rule-based expert system development, multiple-target tracking, orbit determination and Kalman filtering, and time series prediction.

Perlovsky, over the course of his career, has had a great deal of involvement in the development of this research in both his consulting work and his work at Nichols Research Corporation. I know a lot about this because, in 1980 I began working at the Aerospace Corporation in El Segundo California as a MTS (statistician). I worked on statistical problems including Kalman filtering, image processing, image recognition, orbit determination, rule-based expert systems, multiple-target tracking and target discrimination. When I moved over to manage the tracking and discrimination algorithm development for the Air Force's Space Surveillance and Tracking System I became familar with the work being done for us by contractors that included Nichols Research Corporation (NRC). I became very familar with the work of Perlovsky and his colleagues at Nichols Research who supported us from the Newport Beach, Colorado Springs and Boston offices of the company on the multiple-target tracking algorithms and the target classification algorithms. I found the work to be so interesting and of such high quality that I joined NRC in 1988.

Perlovsky sees artifical intelligence as a very practical discipline and believes that computing machines can do a good job of at least mimicking human intelligence through the use of a priori knowledge (as rules preprogrammed into the computer or a priori probability distributions) along with experience (collected data from observational or statistical designed studies) combined using algorithms (Bayes theorem, adaptive neural networks) based on the mathematical foundation of uncertainty incorporated through probability theory and/or fuzzy set theory.

In my experience, I have found rule-based expert systems to be one of the major successful developments in the field of artificial intelligence. At the heart of these systems lies the tools of mathematics and statistics, including the discriminant or classification algorithms based on multivariate Gaussian models (linear and quadratic classifiers) and the nonparametric classification algorithms (kernel discriminant algorithms and classification tree algorithms). Also, patterns can be discovered by computers through the use of clustering algorithms based on Gaussian mixture models or nonparametric techniques like nearest neighbor rules.

The Bayesian approach to statistical analysis has been useful in many areas including the Kalman filter. In Kalman filtering prior knowledge plus current data is used to update the estimate of the current state and for the prediction of the future state of a dynamic system using a simple recursive algorithm that is easily updated in the computer. Many of these developments are well characterized and developed from first principles in this text.

Perlovsky emphasizes his own work including the MLANS system which is a neural network system that incorporates important statistical ideas such as maximum likelihood, the Cramer-Rao inequality and statistical efficiency along with the neural network architecture.

Some of this work was developed by Perlovsky under an Army contract that was coincidentally managed by my brother Julian. I have always viewed this research as being successful because it applied appropriate statistical models to the real problems. I think the crucial aspects of this work are the appropriate use of the Bayesian paradigm and teh indentification of appropriate models for construction of the likelihood equations. The fundamental and well established tools of probability and statistics are the keys. In his proposals, Leonid also included ideas from fuzzy set theory and embedded his methods in an artificial neural network framework. I always thought that these modern theories (fuzzy set theory and neural networks) were gimmicks to get military funding. This may not have been a fair assessment on my part as a careful reading of this book indicates that Perlovsky honestly views these tools as important.

There are subtleties to concepts such as fuzzy set theory. Although I do not yet see its value as a substitute for measure theoretic probability theory for characterizing human uncertainty, it is possible that I just haven't thought hard enough about it. Maybe a continued reading and rereading of Perlovsky's book will help me.

This is a very interesting and unique book on artificial intelligence from a perspective that is quite different from what one find in the standard books written by computer scientists (who often do not have the deep understanding of probability and statistics that Perlovsky possesses).

Perlovsky has published a large number of papers exploring limitations associated with today's main approaches to AI. This book captures the essence of those papers with the addition of developing and expanding Perlovsky's philosophy of AI.
Perlovsky takes the reader through the development of AI from its behaviouristic beginnings through Minsky's "revolution" of purely symbolic AI. He then goes through some more recent methods of combining both methods with some adaptive learning techniques. For all cases Perlovsky clearly demonstrates the inherent limitations of all methods through analytic means.
He then presents his view on a possible way forward though adaptive networks employing fuzzy logic, illustrated with some examples such as work done on SAR image analysis. Throughout the book he provides many examples and certainly this will make an excellent advanced textbook for the field of AI.
I am particularly impressed by his good overview and development of his philosophical views. After years of books by people like the Churchland's, Chalmers, Searle and Hofstadter this is finally a great example of someone who is not afraid to cut through the fluff and expose the real problem to further progress. This should be required reading for anyone looking into philosophy, in particular the philosophy of mind and science. The references are very valuable and Perlovsky and done an excellent job of listing many.
That said, there are a few points I would suggest for the next edition of this book. First, I find it odd that Perlovsky seems unaware of Chaitin's work in algorithmic information theory, particularly his book "The Limits of Mathematics". Considering Perlovsky's many references to Godel and Turing this is a glaring omission. Also, he is missing a reference to Wilson's excellent "Spikes, Decisions, and Actions : The Dynamical Foundations of Neurosciences", possibly because it is also a very recent publication. These three books belong together!
Perlovsky also needs to answer the issues posed by Susan Haack ("Deviant Logic, Fuzzy Logic : Beyond the Formalism") regarding the viability of fuzzy logic and better-define the methods that are the "correct" interpretation when applied to neural networks. I would hope to see an expansion on this theme in the future.
There are some irritating small problems, the most major being the spelling of "Plank" throughout the text; someone's spellcheck was acting up I guess...The index is a bit thin as well.
An excellent book and, as stated in the other reveiw, "must" reading for anyone interested in the future of AI. Perlovsky, Chaitin, Prigogine, Wilson and James H. Austin ("Zen and the Brain") are to be applauded for breaking with the current strangling hold of ancient thought.

The book offers a fresh view of artificial intelligence (AI). Particularly interesting is the drawn analogy between longstanding philosophical questions and problems of modern AI. Even though the knowledge of modern statistical and math concepts is required to understand technical issues most of the ideas are explained on the intuitive level. The key technical problem discussed in this book is multiple model identification from overlapping data samples. The problem is related to computer-based vision, radar target tracking, financial market analysis, and speech recognition. The combinatorial explosion of computational complexity usually is associated with this class of problems and present the great challenge for modern computational science. The author reviews contemporary mathematical approaches, argues that the problem is related to differences between Plato and Aristotle about origin of human intelligence, and explores various approaches among works of famous philosophers in human history. Finally a non-combinatorial solution is offered through the introduction of novel Modeling Field Theory, that combines fuzzy logic with learning and adaptation. This book is a must read for people working with AI.