Design Feature Selection and Classifiers for Hybrid Feature Selection - Particle Swarm Optimization (HFS-PSO)
Kumar Siddamallappa, Nisarg Gandhewar · 2022 4th International Conference on Inventive Research in Computing Applications (ICIRCA) · 2022
In image classification, the feature selection process is critical for reducing overhead and improving accuracy. In light of the increasing growth of data, the concept of integrating classifiers with feature selection approaches has lately emerged as a new trend for boosting the performance of classification systems. There are several benchmark algorithms that use wrapper, filter, or hybrid approaches. These algorithms use a variety of strategies, ranging from basic search -based totally strategies to more complex nature-stimulated set of rules-based totally strategies. In this paper, a hybrid feature selection set of rules based on image classification is proposed. The suggested framework's performance was compared to many benchmark supervised and unsupervised feature choice techniques and found to be advanced. Moreover, the advised algorithm is much less computationally extensive and therefore took much less time to execute for the publicly available datasets used within the trials, which covered high-dimensional datasets.