Algorithms Books

Buckley and Feuring have written a concise and understandable presentation of the essential mathematics involved in artificial neural nets (fuzzy and conventional), in fuzzy expert systems, and in hybrid systems employing both techniques. The book is most valuable for those seriously interested in such systems whose mathematics is primarily of the engineering variety. A real virtue is that the options available in the design phase are spelled out; it is not assumed that the system designer will follow the conventional well-trodden path. The example given of a fuzzy teaching machine illustrates choosing among these options. The treatment of neural nets is especially good; I would wish that the treatment of fuzzy expert systems had been expanded beyond bare necessity. Useful adjunct reading is Earl Cox's Fuzzy Systems Handbook and von Altrock's Fuzzy Logic and NeuroFuzzy Applications.

While much of the basic mathematics is available in other books such as Klir and Yuan, it is quite difficult to locate what one needs to know in these books. Buckley and Feuring's presentation has the great virtue that the math is organized in a fashion appropriate to the system designer's needs. I know of no other book which fills this niche.

This book contains a very thorough and concrete presenation of combined soft computing and image processing approaches. Even though a relatively wide knowledge of fuzzy logic is required, this book can be thought of as "the best of both worlds".
The chapter "Fuzzy data fusion for multiple cue image and video segmentation", by Ioannou et al., is a representative example of the book's completeness, since it describes fuzzy data fusion techniques capable of integrating multicue segmentation results to provide consistent image partitioning based on moving objects. I'm looking forward to more articles by these authors, since their style of writing is clear and concise. Good work, authors and editors!

Fuzzy sets based heuristics for optimization provide a bridge between the real world and theoretical concepts an intelligent artificial system. It has become an important tool for different engineering applications. This book will appeal to senior undergraduate and graduate students in most engineering disciplines and academics, as well as some working in economics, control theory, operational research, engineering for industrial production, etc.

Some contributions follow:

- Fuzzy Adaptive Neighborhood Search called FANS, fuzzy set based heuristic to improve the quality of life of human beings, with didactic examples of optimization applications. I reckon that in a few years FANS will be employed in the optimization of resources in the trade and industry fields.

- In the logistic field there are three very interesting studies to conjugate trajectories (trajectories shorter, quicker, more reliable, etc.) Vehicle Routing Problem with Uncertain Demand at Nodes: The Bee System and fuzzy Logic Approach, b) Capacitated Vehicle Routing Problem with Fuzzy Demand, Route Choice Making Under Uncertainty: a Fuzzy Logic Based Approach and c) Multi-stage Supply Chain Network by Hybrid Genetic Algorithms.

- The contribution Fuzzy Constructive Heuristics from Heuristic Constructions would be good to replace the present techniques of comparison and estimation, called evaluated times; for example, in the calculations of costs in the cloth cuts. This would reduce the waste.

- Fuzzy Evolutionary Approach for Multiobjective Combinatorial Optimization, presents an interactive algorithm for multiobjective fuzzy programming problems with the preference level from Pareto optimal solution to allow a formulation, with restrictions, of water quality and economic deal for industrial pollution management.

- Very useful for operation planning techniques of district heating and cooling is the contribution whose title is: An Interactive Fuzzy Satisfycing Method for Multiobjective Operation Planning in District Heating and Cooling Plants through Genetic Algorithms for Nonlinear 0-1 Programming.

Many papers in many different topics related to evolutionary computation. The conference is organized around independent "demes" with the goal of ensuring diversity and respect for emerging ideas. The standards for reviewing are relatively high. However, there are some papers that obviously fell though the cracks, which is probably inevitable in a conference of this size.

I attended this conference in San Francisco and I learned many things in many different areas of evolutionary computation. If you want to learn about the latest in this field, get this collection of good papers. All the papers were reviewed and the conference has a moderate acceptance rate, so there are very few disappointing papers.... I think it is worth it.

Reeves and Rowe provide a very readable overview of what is currently known about evolutionary algorithms. While the book provides very good coverage of several recent models of GA dynamics, it would be worth the money if only for the chapter on the dynamical systems treatment of GAs.

Vose's "The Simple Genetic Algorithm" is a brilliant piece of work, but it also holds the distinction of being the earliest in a book that I have ever become utterly, completely, hopelessly lost. I refer to this book as "the one I use to figure out what Vose was talking about."

This was both a good introduction to genetic algorithms in general and to their adaptation to grouping problems in particular.

More than that the style is concise, engaging and consistently interesting. The explanations are practical and intuitively appealing.

I found myself jumping around in the book somewhat (I would recommend reading chapter 14 before chapter 1), but everything was explained somewhere. At the end, I felt I had a good basis not just from which to implement a grouping GA, but to consider many other variations or developments from the classic model.

The key strength of this book for me was its practical, common-sense approach that stresses understanding the problem and if it works use it, if not adapt it so it does.

Rather the spirit of GAs themselves.

Why be interested in geometric numerical integrators you ask? Well, if you want algorithms that are accurate and conserve geometric properties of the dynamical flow you have to look at geometric integrators. For instance, in Hamiltonian systems the flow preserves the symplectic two-form (the volume of the phase space), and symplectic integrators do exactly that, which means that these algorithms are more accurate than their non-symplectic counterparts: conserved quantities are really conserved numerically.

Geometric Numerical Integration deals with the foundations, examples and actual applications of geometric integrators in various fields of research, and there is a lot on the more abstract theory of numerical mathematics, the classification of algorithms, provided with lots of mathematical and physical background needed to understand what is special about certain algorithms and advice on when, where and how to use them. It is completely self-contained, up-to-date, clear, well written, it has many references, and it is aimed at students and scientist who want to learn more about everything there is to know on geometric integrators.

Admittedly, it is not completely inexpensive, but considering it is probably the only book you'll ever have to buy on geometric numerical integration and the fact that it looks great and is made very well, it is well worth the money!

I took the graph theory course in a semester when Shimon Even lectured himself, using this book as the course's text book.

Of the lecturers in the Technion, Shimon Even is one of the best, and the book reflects this. The material is clearly presented in a very understandable matter, and requires no further references (beyond the required discrete math).

I'm still using material from the book (and course) in my work, as I prefer it over many other newer books.

It's sad that the book is out of print, but a couple of students have scanned the first six chapters and posted it in postscript format, and made it available [online]

This is a very good introductory book on Graph Theory. If you don't want to be overwhelmed by Doug West's, etc., and yet receive a decent introduction to the topic, this book is your best bet. It covers all the topics required for an advanced undergrad course or a graduate level graph theory course for Math, engineering, operations research or computer science students in good depth and details. There are good examples and interesting exercises; some computer codes (JAVA) are also available in the book implementing some of the algorithms.
I would say O.R. and CS people will benefit a lot from it both as a reference or a textbook if adapted for a one semester graduate course. The only drawback is the price!