Genetic algorithms applied to multi-class

prediction for the analysis of gene expression data

Mar. 05,2003

Chia-ming Chang

Abstract :

              The team (The Omniarray Group is based at the National Cancer Centre of Singapore.) described genetic algorithms (GAs) based method to predict the multi-class with gene expression data. The advantage of the GA algorithm is automatically determines the members of a predictive gene group that maximizes classification success using a maximum likelihood (MLHD) classification method. The group believe their method is better than other published methods using the same dataset.

               GAs (genetic algorithms)  developed by J. H. Holland in 1975 that manipulates bit strings analogously to DNA in evolution. The "organisms" in genetic algorithms are termed individuals. Genetic algorithms are stochastic approaches based on the concept of biological evolution and biological genetics. Genetic algorithms operate on chromosome-like data structures that encode possible solutions of the problems, and apply crossover and mutation operators to generate new chromosomes in a search space. Then, based on the principle of survival-of-the-fittest, chromosomes with good performance are selected though selection operator.

(http://www.omniarray.com/bioinformatics/GA )

Source:

from: Bioinformatics 19(1): 37-44, 2003  

Reference:

Scott,A., 1990. An Introduction to Genetic Algorithms. AI Expert, Vol.4, No.3, pp.49-53.

 

K. D. Jong, 1988. Learning with genetic algorithms: An overview. In Machine Learning , No.3, pp.121-138.

Yeang,C.H., Ramaswamy,S., Tamayo,P., Mukherjee,S., Rifkin,R.M., Angelo,M., Reich,M., Lander,E.S., Mesirov,J.P. and Golub,T.R. 2001 Molecular classification of multiple tumor types. Bioinformatics, 17, S316–S322.