An efficient DBSCAN optimized by arithmetic optimization algorithm with opposition-based learning

Abstract As unsupervised learning algorithm, clustering algorithm is widely used in data processing field. Density-based spatial of applications with noise (DBSCAN), as a common can achieve clusters via finding high-density areas separated by low-density based on cluster density. Different from other methods, DBSCAN work well for any shape the database and effectively exceptional data. However, employment DBSCAN, parameters, EPS MinPts, need to be preset different object, which greatly influences performance DBSCAN. To automatic optimization parameters improve we proposed an improved optimized arithmetic (AOA) opposition-based (OBL) named OBLAOA-DBSCAN. In details, reverse search capability OBL added AOA obtaining proper adaptive parameter optimization. addition, our OBLAOA optimizer compared standard several latest meta heuristic algorithms 8 benchmark functions CEC2021, validates exploration improvement OBL. validate OBLAOA-DBSCAN, 5 classical methods 10 real datasets are chosen compare models according computational cost accuracy. Based experimental results, obtain two conclusions: (1) OBLAOA-DBSCAN provide highly accurately more efficiently; (2) significantly ability, better optimal parameters.