Genetic algorithm optimized feature transformation: a comparison with different classifiers

摘要: When using a Genetic Algorithm (GA) to optimize the feature space of pattern classification problems, performance improvement is not only determined by data set used, but also depends on classifier. This work compares improvements achieved GA-optimized transformations several simple classifiers. Some traditional transformation techniques, such as Principal Components Analysis (PCA) and Linear Discriminant (LDA) are tested see their effects GA optimization. The results based some real-world five benchmark sets from UCI repository show that after in reverse ratio with original rate if classifier used alone. It shown performing PCA LDA prior optimization improved final result.