An efficient gene selection method for microarray data based on LASSO and BPSO

摘要: The main goal of successful gene selection for microarray data is to find compact and predictive subsets which could improve the accuracy. Though a large pool available methods exists, selecting optimal subset accurate classification still very challenging diagnosis treatment cancer. To obtain most genes without filtering out critical genes, method based on least absolute shrinkage operator (LASSO) an improved binary particle swarm optimization (BPSO) proposed in this paper. avoid overfitting LASSO, initial divided into clusters their structure. LASSO then employed select high further calculate contribution value indicates genes’ sensitivity samples’ classes. With second-level established by double filter strategy, BPSO encoding information obtained from perform selection. Moreover, perspective bit change probability, new mapping function defined guide updating more BPSO. strategies, with probability. experimental results several public extreme learning machine verify effectiveness compared relevant methods.