Comparative analysis of genetic algorithm and particle swam optimization: An application in precision agriculture
Asian Online Journals
School of Science
This article details the exploration and application of Genetic Algorithm (GA) and Particle Swam Optimization (PSO) for the wrapped-based feature selection. Particularly a comparative study is carried out, examining the performances of both GA and PSO with respect to classification accuracy of some classifiers. 112 features were extracted features from set of images found in the Flavia dataset (a publicly available dataset). The extracted features are Zernike Moments (ZM), Fourier Descriptors (FD), Legendre Moments (LM), Hu's Moments (Hu7M), Texture Properties (TP), Geometrical Properties (GP), and Colour features (CF). The main contribution of this article includes the comparison of two major optimization techniques, i.e., GA and PSO, and the development of a GA-based feature selector using a novel fitness function which enabled the GA to obtain a combinatorial set of feature giving rise to optimal accuracy. The effectiveness of these manifold projection techniques were tested on Probabilistic Neural Networks (PNN), k Nearest Neighbour (kNN) and Multilayer Perceptron (MLP). The experimental analysis demonstrates the classification accuracy with GA-based approach outperforming that with PSO-based method.