A Modified Fuzzy K-nearest Neighbor Using the Improved Sparrow Search Algorithm for Two-classes and Multi-classes Datasets

Abstract

The Sparrow search algorithm is a new and effective swarm intelligence method proposed in recent years and studied in many publications. Based on the basic principle of sparrow search algorithm, this paper combines the inverse learning algorithm with the refined inverse solution to form an improved sparrow search (SSA) algorithm. Combining the fuzzy k-nearest neighbor method and the improved SSA, the numerical simulation of two-classes datasets and multi-classes datasets is carried out, and many numerical results are obtained, and the results are analyzed. At the same time, this paper lists the data comparison results and tables with other models. The hybrid SSA-FKNN proposed in this paper has a clear advantage in terms of accuracy (ACC).