Enhanced Accuracy Enabled by Particle Swarm Optimization in Classification Application
Zhiyi Lu · 2020
Particle Swarm Optimization (PSO) is a stochastic optimization algorithm which has great potential in producing better accuracy results related to parameter settings. PSO with other mathematical models shows great potential in improving quality and accuracy. Researchers have been using this algorithm in multiple fields related to machine learning. This paper aims to analyze PSO's great potential in optimizing models used in classification, showing its effectiveness and suggests further researches on this algorithm. Support Vector Machines(SVM) and Convolutional Neural Networks(CNN) are common algorithms used in Image Classification. Thus, this paper analyzes research done on PSO-SVM and PSO-CNN, in order to differentiate their advantages and disadvantages in Image Classification. The author also discusses overfitting and solutions to avoid overfitting of both algorithms. While using PSO to optimize the combined SVM-CNN model, the result outperformed all conventional models. This outcome illustrates PSO optimized SVM-CNN might be the best model to use in classification problems, and suggests the effectiveness of PSO in optimizing models related to classification.