A Memetic Cellular Genetic Algorithm for Cancer Data Microarray Feature Selection

MATIAS GABRIEL ROJAS; OLIVERA ANA CAROLINA; CARBALLIDO JESSICA A.; PABLO VIDAL

IEEE LATIN AMERICA TRANSACTIONS

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC

Lugar: New York; Año: 2020 vol. 18

1548-0992

Gene selection aims at identifying a -small- subset ofinformative genes from the initial data to obtain high predictiveaccuracy for classification in human cancers. Gene selection canbe considered as a combinatorial search problem and thus canbe conveniently handled with optimization methods. This paperproposes a Memetic Cellular Genetic Algorithm (MCGA) to solvethe Feature Selection problem of cancer microarray datasets.Benchmark gene expression datasets, i.e., colon, lymphoma, andleukaemia available in the literature were used for experimentation.MCGA is compared with other well-known metaheuristic?strategies. The results demonstrate that our proposal can provideefficient solutions to find a minimal subset of the genes.