Artificial Intelligence Books

The author, James Shanahan is internationally renowned across a number of computing fields including AI, information retrieval, information access, linguistics, text/data mining and machine learning. Indeed his expertise is reflected in his already substantial and growing IP portfolio. The book quickly made it's mark leading to a substantial international speaking itinerary and it has directly influenced many new approaches, research and developments in soft computing and knowledge discovery.

The book forms a comprehensive discussion and overview of the traditional soft computing approaches to knowledge discovery and, importantly, advances the science to Cartesian granule features and their corresponding learning algorithms thus introducing an intuitive approach to knowledge discovery.

The book and it's contents are well thought out both in how it presents complexities and in how it argues and presents this new approach vis a vi the traditional and other approaches. The approach is supported with examples applied to diverse scenarios including object recognition in outdoor scenes and in medical diagnosis and control.

It's not just a book but an essential reference source, teaching tool and resource for all with an interest in the topic. It provides a host of datasets and source codes for several algorithms described in the book.

Don't stop with this book. James Shanahan continues to surprise and amaze in the computing field and can be found presenting at numerous events each year from Asia to the US to Europe. He has an avid following and is certainly `one to watch'.

I eagerly waited this book, and was not dissappointed. This book covers in 15 chapters one of the hottest topics in computer advancements - intellegent agents and their important role in future comm based systems. The first two chapters are an introduction to the subject, begining with the low level details, so each can find his/her way and feel right at home. The rest 13 chapters introduce different problems by various groups of authors and researchers and their approach to the presented problem. This book is an eye opener and provocator. I like the diversity and approch, as well as always having the pleasure of reading such excellent work. This is not a technical book, and as such is a great starting point to the exciting area of intelligent agents. If you are starting your way, or would like to have a good look around this is definately the book for you. If you want a more technical book, the "Multiagent Systems : A Modern Approach to Distributed Artificial Intelligence" by Gerhard Weiss is a better resource. But All in all, I love this book!

There seem to be many robot/AI books out there but most try to cover too much. I am impressed with how well written this book is considering the complex concepts it tackles.

Definately not for casual readers; you don't need a strong AI background but you will not enjoy the book if you don't want to delve into the esoteric concepts of representing space and time.

I read the first edition of that book and it is terrific. The second edition is much more adapted to current research. Statistical methods in NLP are more detailed and some syntax-based approaches are presented. My specific interest is in machine translation and dialogue systems. Both chapters are extensively rewritten and much more elaborated. I believe this book is perfect for everyone who starts in speech and language processing. With precision, coherent examples and some humor, this book give a great introduction into this topic as well as material for already experienced readers.

Siciliano and Khatib have assembled a massive and comprehensive tome on robotics, circa 2008. Sections of the book can be read by a diverse audience of undergraduate and graduate students, researchers and even the general public. Spanning any field associated with the subject.

There is considerable maths in the modelling of robots. Often to understand and control an arm. The multiple degrees of freedom of joints are wonderful for dexterity. But these often give an excursion into advanced linear algebra and control systems theory. Several chapters go into the necessary maths. You probably need at least 2 years of undergraduate engineering maths as preparation.

The myriad applications in which robots have been deployed is amply surveyed in Part F, Field and Service Robotics. In the household, there is of course the floor cleaning Roomba. A cute little gizmo, but it is not a toy; a genuine robot in its own right. The chapter mentioning it also describes an entire genre of competitors; mostly lesser known to the public.

Another chapter on agriculture and forestry talks about using robots for tasks like harvesting. Usually more successful when the terrain is flat and well defined; ie. having only one crop present. While the general case of a robot in hilly, wooded terrain with multiple obstacles and different species of trees is much harder to program.

I also ran into something in this chapter from my past, and it impressed me as to the book's comprehensiveness. At the University of Western Australia, there was a long running program to devise a robot sheep shearer. It started in the 70s and I met several of its researchers. I lost track of it after 1983, but I'd wondered whatever became of it. The book takes up the thread, explaining that the program took on the name Shear Magic, and was ultimately discontinued because it was never fast enough. But even in failure, this robotic application had a side effect. The demonstration of the technology was used by farmers to browbeat human shearers into moderating their wage claims, by playing off longstanding fears of workers about being replaced by machines. Of course, whether or not this was desirable may be a function of your political leanings.

To me, the most interesting section of the entire book concerned mirror neurons. This was a fundamental recent discovery in biology. The relevance to robotics is still perhaps speculative. Several robotics researchers have attempted to use it as inspiration for teaching a robot via its visual input and processing system. This contrasts greatly with the traditional teaching use of rule based formal logic, often involving the predicate calculus. The results described in the text are early but promising.

One slight curiosity is the relative deprecating of military applications. These are numerous and scattered throughout various chapters. Covering uses like landmine detectors, or the aerial Predator and its relatives that have seen much recent use in Iraq and Afghanistan for surveillance and attack. But at the top level of the Contents, there is no section on the military. And if you go to the Index, "military" is absent, while, for example, "mind reading" gets 2 entries. The downplaying of the military is especially puzzling given the historically prominent role of the US military in funding advanced robotics research.

I may never finish a proper review. So let me at least present the Table of Contents and a quick guide for Philosophers of Science. The non-technical chapters are rich with new (or uncommon) insights about Induction, Explanation, Theory Choice, and even the Arrow of Time.

Table of Contents
------------------------------------------
1. Inductive Inference
2. Information
3. Strict Minimum Message Length (SMML)
4. Approximations to SMML
5. MML: Quadratic Approximations to SMML
6. MML Details in Some Interesting Cases
7. Structural Models
8. The Feathers on the Arrow of Time
9. MML as a Descriptive Theory
10. Related Word

Philosophers of Science should read:
-------------------------------------
* Chapter 1, at least through 1.5 (18 pages).
- - MML as a precise rendering of Occam's Razor
- - Why the best explanation is the shortest
- - MML as induction (pure Bayesianism as deductive)
- - MML explanations, induction, unification, etc.
* Chapter 9 (12 pages)
- - The big picture: can you use MML to describe scientific revolutions?

Those interested in the Arrow of Time should read
---------------------------------------------------------
* Chapter 8 (38 pages)
- - A very careful account of reversibility and irreversibility.
- - Accurate simulations to convince people not already convinced of Boltzmann's claim that entropy will increase in both directions.
- - A novel account of asymmetry, suggesting that while we predict the future, we EXPLAIN the past, in the MML sense. That is, MML inference naturally picks out the past we remember, as it is the best explanation of the present.

Anyone wanting more details should begin with:
----------------------------------------------
* Chapter 2 (87 pages)
- - Shannon Information, coding, and entropy
- - Algorithmic Complexity
- - Information, Inference, and Explanation

The Travelling Salesman Problem (more politically correctly called the Travelling Salesperson Problem)(How about we call it the TSP?) is a common real world problem. The problem is simplely stated: How do you find the shortest path for a travelling sales_____ to drive as he visits a series of customers.

If you're just running a few errands it's easy enough. If you're going to the supermarket, the post office, the dry cleaners, and the gas station, it's pretty easy to determine the shortest path. But then you add constraints, nearly out of gas, go to gas station first. Buying ice cream at the super market, better make it the last stop.

This is a real world problem. It is faced every day by delivery companies like the post office and air freight companies who spend huge amounts of fuel flying jet freighters around the world -- where do you put a hub, how do you schedule everything to come together.

FedEx solved this problem by running everything through Memphis. This works really well for a package going from New York to LA. It works less well for a package going from Manhattan to the Bronx. And if the package is going from London to Manchester ....

The computation of this and similar problems fall into the general category of Stochastic Local Search. And this book is the first to offer a systematic and unified treatment of SLS. Before this there were a series of technical papers, magazine articles and chapters in more general texts. The book provides the first unified view of the entire field and offers an extensive review of state-of-the-art algorithms and their applications. A companion website offers lecture slides as well as source code and Java applets for exploring and demonstrating the algorithms.

You'll find this book useful if you ever need to analyze, interpret, or categorize strategies. It ties together strategies across different domains and shows the commonalities between them. It is a good source of ideas for further work in this new field...as the coverage is broad, it gives a view of the lie of the land without making too many claims. It's also a very portable book, kind of like a book on trees/stars/birds that you might take on a camping trip.

I found the beginning of the book a bit tedious but, once I made it past the background material and into the core concepts, it got really exciting. This book proposes a technique for dynamically evolving the structure of a nueral net during training. I haven't tried implementing it, but it's a fascinating theory.

How does a three-and-half gelatinous blob -- the human brain --give rise to consciousness? Our teetering bulbs of dread and dream, as the poet Russell Edson once described the brain, are the most complex objects in the known universe. But despite the vast power of the cerebral cortex we still do not understand that most implacable of mysteries, our neurologically (and bodily) based selves. Laura Weed in her vastly edifying "The Structure of Thinking" sheds an enormous amount of light into the dark recesses of the human mind. Weed's book is one of the best accounts of what thought is, and what it is not. Thought, according to Weed, isn't reducible to syntax, thought is not just an algorithm or computer program, the crunching of abstract symbols as part of some sort of neuronally based computation. Thought requires an 'intentional' aspect, a point of view, and our neural systems have evolved to interpret, categorize, or mark out aspects of our experience in specific ways, ways that are not reducible to computation. Weed provides some very creative and original ways of reassessing our approach to consciousness, including a very promising discussion on 'intentional causation' or what Weed dubs 'Kausation' as a way of overcoming difficulties associated with the notion of 'efficient causation' which dominates most thinking in the cognitive sciences. My summary here is necessarily an over simplification of a rich and very subtle book. But for anyone interested in a post-Cartesian process account of mind "The Structure of Thinking" is a uniquely rewarding book.