GENETIC ALGORITHM BASED INSTANCE SELECTION FOR NEAREST NEIGHBOR RULE

摘要

Instance selection is an important pre-processing step in pattern recognition and machine learning. In this paper, we propose a novel instance selection method based on genetic algorithm for nearest neighbor (AGAIS_NN), which compose of three main parts: elitist strategy, adaptive probabilities of crossover and mutation, and fitness function. To validate the proposed algorithm, we compare AGAIS_NN with other classical instance selection methods. The experimental results show that our proposal is more effective and useful than other approaches.