On-Line Learning of Write Strategy for Ultra-Speed CD-RW Optical Recorder

An on-line machine learning approach integrating the genetic algorithm (GA) and jitter measurements is proposed to learn the write strategy for the infrared diode of ultra-speed CD-RW recorders. The recording performance differs significantly for the CD-RW discs recorded for the first, second, or third time above. It is difficult to learn one set of write strategy parameters for the infrared diode of ultra-speed CD-RW recorder that satisfies the recording specifications for three different types of discs. The GA is applied to the on-line learning of write strategy. However, the convergence of GA stagnates at the final stage of the learning process due to the fact that the write strategy parameters learned by the GA need to satisfy the recording specifications for discs recorded for the first time, second time and third time within one recording trial. To overcome this difficulty, a scheme called dynamic parameter encoding is proposed. This scheme improves the GA convergence and explores the search space much better than the conventional GA.

Evolutionary Computation and Its Applications in Neural and Fuzzy Systems

Neural networks and fuzzy systems are two soft-computing paradigms for system modelling. Adapting a neural or fuzzy system requires to solve two optimization problems: structural optimization and parametric optimization. Structural optimization is a discrete optimization problem which is very hard to solve using conventional optimization techniques. Parametric optimization can be solved using conventional optimization techniques, but the solution may be easily trapped at a bad local optimum. Evolutionary computation is a general-purpose stochastic global optimization approach under the universally accepted neo-Darwinian paradigm, which is a combination of the classical Darwinian evolutionary theory, the selectionism of Weismann, and the genetics of Mendel. Evolutionary algorithms are a major approach to adaptation and optimization. In this paper, we first introduce evolutionary algorithms with emphasis on genetic algorithms and evolutionary strategies. Other evolutionary algorithms such as genetic programming, evolutionary programming, particle swarm optimization, immune algorithm, and ant colony optimization are also described. Some topics pertaining to evolutionary algorithms are also discussed, and a comparison between evolutionary algorithms and simulated annealing is made. Finally, the application of EAs to the learning of neural networks as well as to the structural and parametric adaptations of fuzzy systems is also detailed.

Detection of closely linked multiple quantitative trait loci using a genetic algorithm

The existence of a quantitative trait locus (QTL) is usually tested using the likelihood of the quantitative trait on the basis of phenotypic character data plus the recombination fraction between QTL and flanking markers. When doing this, the likelihood is calculated for all possible locations on the linkage map. When multiple QTL are suspected close by, it is impractical to calculate the likelihood for all possible combinations of numbers and locations of QTL. Here, we propose a genetic algorithm (GA) for the heuristic solution of this problem. GA can globally search the optimum by improving the "genotype" with alterations called "recombination" and "mutation." The "genotype" of our GA is the number and location of QTL. The "fitness" is a function based on the likelihood plus Akaike's information criterion (AIC), which helps avoid false-positive QTL. A simulation study comparing the new method with existing QTL mapping packages shows the advantage of the new GA. The GA reliably distinguishes multiple QTL located in a single marker interval.