Customer Reviews

Ant Colony Optimization (Bradford Books)

5 star:		(5)
4 star:		(1)
3 star:		(0)
2 star:		(1)
1 star:		(0)

 

 

Being an ant isn't very complex, but it's a daily fight for life. The losers in that fight don't count, but the winners get to vote.

That is the basis of ant colony optimization. There are many parts to the idea, all of them very simple. First, there are many routes to the goal (food, if you're an ant) - some are better, some worse, you don't know which are which in advance, and the answer may change over time. Second, it's a random search. If you find any answer at all, no matter how convoluted, you get to vote on your route. Third, there are many other ants, all voting. Any leg of a trip that is heavily followed must be part of a good route, and gets many votes. There are details, but that's about it.

Chapters 1-3 are the most readable, and convey the basic spirit of the family of algorithms. Ch. 4-6 will drag a bit, for the general reader, but go into significant detail about the ant algorithm and specific applications.

Ch. 7 ends the book with a warm, informal discussion of the algorithm's history and some delightful variations. Dorigo, the principal author and founder of the ant school, uses this chapter to express his pure joy at having found such a wonderful thing, and at the similar approaches that others have also found.

The approach has some real limits. For example, it can solve only problems that look like finding the shortest route. The good news is that a wide range of unlikely problems can all be cast in these terms. The better news is that, given the many variations available, some form of the 'stigmergic' approach will probably solve any problem in that range. Best of all, though, is the sheer cleverness and the sincere appreciation expressed by the authors.

Nature is economical, but a brilliant problem solver. This is written by someone who as able to listen in on one of the lessons.

//wiredweird

Fifteen years after the elegant double-bridge experiments by Deneubourg et al. that formed the basis of the Ant Colony Optimization algorithm, Marco Dorigo, the inventor of ACO, and Thomas Stützle, an expert on stochastic local search methods, have pooled their knowledge to summarize the current state of the art.
This book gives a well paced introduction to ACO, describes its use in various optimization problems and gives interesting examples of its applications in industry. Explanations are clear and concise and, with the exception of a few well defined technical terms, free of scientific jargon. It is a pleasure to read for everyone with an interest in optimization theory. However, if you are looking for a book that celebrates the beauty of nature's problem solving capabilities, you are better of with "Swarm Intelligence" or Flake's "Computational Beauty of Nature". The initial idea of ACO may be bio-inspired, but this book has a crystal clear focus of the computational considerations in optimization theory.

This book is a fine compilation of what have been done with the Ant Colony paradigm so far. Highly readable, even for people without previous experience in the field of optimization.

You have to get to page 99 to find an explanation of how to implement the algorithm. That explanation is much like the preceding material. A meandering outpouring of facts without much regard for consistency or clarity and an unnecessary tendency to over complicate. Caters to the research community at the expense of clarity to the student.

The authors need to explain the algorithm first in clear and simple terms - then they can talk about optimizations!!! There is no point telling me how to optimize something you haven't fully explained.

The consequence of the approach in this book is that finding out useful practical information becomes ordeal by treasure hunt. For example. Try finding an explanation for the derivation of the parameter for pheromome update. There's a nice greek symbol for it and it's between 0 and 1 but er..... thats it.

If these guys are professors I pity their students.

The concept is fascinating and the applicability to numerical optimization is beautifully presented. In terms of presentation and information provided to the reader, this is a perfect combination of motivation and insight. The potential of the method in solving complex problems is a different story and needs further investigation.

Ant Colony Optimization focuses on the fact that ants foraging for food will quickly form a trail that is the shortest possible ditance betwen the food and home. Rach ant follows the scent trail laid on a path by previous travelers and adds its own pheromone to the scent, both going and coming. With a choice, ants tend to follow the strongest scent. Of a pioneer pair, the one choosing the shortest path will make the round trip before the other. Each pheromone trace evaporates in time, but an ant's antenna can detetct the slightest trace. That is a simplification of the introductory chapters of the book. The "pheromone trail" scheme is used to devise "artificial ant" which then takes part in the comnstruction of powerful ant algorithms for solving intractable problems such as the classical "Traveling Salesman" and other routing problems. The book is a complete text for a college course, with a large bibliography and many internal references to sources on the Internet. It is well written, with pseudo code showing how each algorithm can form computer programs. I can't evaluate the difficulty, but for me the math in later chapters is above my reach, but gratifying, nevertheless.

The central idea in the book is to analyse what evolution has provided us. In the form of ants being able to find the shortest path over terrain. This ability has inspired the research described herein.

The book can be read as a fascinating deconstructionist approach to observing and manipulating ant colonies. By trying to look under the observations to discern the fundamental algorithms at work. And then to apply these to such longstanding contexts as the Travelling Salesman Problem.