A Hybrid Feature Selection Algorithm Based on Collision Principle and Adaptability

A Hybrid Feature Selection Algorithm Based on Collision Principle and Adaptability

Blog Article

Feature selection plays a significant role in machine learning and data mining, where the goal is to screen out the most representative and relevant subset of features from a large collection of features to improve the performance and generalization ability of the model.In this paper, a hybrid feature selection algorithm that combines a filter algorithm and an improved particle swarm optimization algorithm is proposed, that is, the Information Gain and Maximum Pearson Minimum Mutual Information improved Adaptive Particle Swarm Optimization algorithm (IGMPMMIAPSO).First, transpharm online shopping combined with the characteristics of the Pearson correlation coefficient and mutual information, a filter algorithm called Maximum Pearson Minimum Mutual Information (MPMMI) is proposed.The algorithm balances the relevance and redundancy between the features by adjusting two weight parameters ( $w_{p1}$ and $w_{p2}$ ).

Second, Adaptive Adjustment of Control (AAC) is introduced to update the particle swarm optimization algorithm, so that the particle velocity has a higher searching ability, and the diversity of population position changes is increased.The improved algorithm was used as the wrapper algorithm.Simultaneously, the concepts of the No Continuous Change (NCC) times and collision distance values are proposed.According to these, the IGMPMMIAPSO algorithm is proposed by combining the filter algorithm and wrapper algorithm.

To verify the performance of the proposed algorithm, we experimented with other state-of-the-art hybrid algorithms using eight datasets.The truvisionhealthftp.com experimental results show that the classification accuracy of the proposed algorithm is at least 0.1% higher than that of the other five algorithms, and the feature subset length is shorter.