Summary of the review by G. L. Pappa in GPEM, published online on November 25, 2009.
DOI 10.1007/s10710-009-9095-0 © Springer Science+Business Media, LLC 2009.
The review starts with a description of the structure of the book; it is mentioned that "the authors introduce the theory and show extensive experimental results and their analysis for enhanced replacement strategies".
The review mentions the authors elaborations on the combination of "essential" genetic information of individuals as well as offspring selection (OS) and the Relevant Alleles Preserving Genetic Algorithm (RAPGA); OS is described as a method that guarantees that the next population shall always contain individuals that are better than their parents, and the RAPGA is summarized as a combination of "offspring selection with a strategy that allows only individuals having new alleles (not yet represented in the population) to be included in the new population". In the review, this part is described as the core of the book.
Subsequently, a parallel genetic algorithms using OS, called Self-Adaptive Segregative Genetic Algorithm including aspects of Simulated Annealing (SASEGASA) is described; it is described as a method that "is based on the idea that splitting the original population into subpopulations broadens the search. Moreover, its subpopulations can grow in size or disappear over time. A subpopulation becomes extinct when it prematurely converges."
Following the theoretical part, the empirical part of the book is summarized. For instance, approaches for measuring an individual's genetic contribution to the next population are descibed, and also methods for measuring the genetic diversity in GAs (single as well as multi-population algorithms). The reviewer also refers to the author's tests using the TSP to illustrate the main effects on evolution introduced by OS and the RAPGA. As mentioned in the review, these experiments show that essential alleles are preserved when using OS or the RAPGA; this "validates the building blocks hypothesis, and also suggests the new algorithms do not need to rely on mutation as much as traditional GAs".
Furthermore, the authors' tests are described in more detail. On the one hand, the tests using the TSP and the vehicle routing problem (VRP) as well as its derivatives (namely the capacitated VRP and the capacitated VRP with time windows (CVRPTW)) are mentioned; on the other hand, test series applying GP to data-based modeling are also alluded. The HeuristicLab, the authors' framework for heuristic optimization is also mentioned.
The summary of the test series published in this book is also includes remarks on tests using OS and the SASEGASA for solving problem instances of the TSP and CVRPTW as well as real-world applications of GP, for example the design of virtual sensors for emissions of diesel engines as well as classification models for medical datasets.
The final part of the review starts with the remark that the book covers basics of GAs and GP and presents "results of interesting research on new methods capable of preserving relevant building blocks". Finally, the book is called a "good buy for students and researchers interested in genetic algorithms and their application in optimization problems".