Research Of Restraint Strategy For GEP Prematurity

Data mining has already been a hot issue in research and application of database realm. Among various data mining technologies, function mining is an important and promising branch. Evolutionary Computing is often used for function mining. Among various evolution models, Gene Expression Programming, which combines the simple structure of GA and the ability to solve complex problems of GP, maintain simple expression and high adaptability. However, traditional GEP algorithm may falls short in a local optimum pitfall known as premature convergence problemIn order to solve this problem, this paper has done the following work:(1) Analyzes the prematurity of GEP and demonstrate its side-effect on GEP function mining.(2) Proposes a novel GEP method based on backtracking referring to the atavism in the biosphere. Proposes the concept of Backtracking Checkpoint and designing Geometric Proportion Increased Checkpoint Sequence and Accelerated Increased Checkpoint Sequence to restrict the backtracking process.(3) Extends the GEPBS algorithm and for control strategy: (a) Retrogression Factor (RF) Strategy; (b) GEP with proportional backtracking strategy( GEPPBS); (c) Self-adaptive backtracking strategy (GEPSBS); (d) Pruning strategy. This strategy is based on the pruning method in artificial cultivation into traditional GEP algorithm and to...