BGRO: A Binary Golden Ratio Optimization Algorithm for Wrapper-based Feature Selection

Data sets have grown in size and complexity as a result of recent advances in science, engineering, and technology. As a result, machine learning techniques and data mining techniques can't efficiently utilize these large datasets due to their redundant, noisy, and irrelevant features. Through feature selection, datasets can be reduced in dimensionality while increasing classification accuracy by identifying the most relevant attributes. In recent years, meta-heuristic optimization techniques have gained popularity as a means of feature selection because they can overcome traditional optimization limitations. This paper presents a binary version of the Golden Ratio Optimizer (GRO) as Binary Golden Ratio Optimization (BGRO), an alternative optimization algorithm. Using five standard UCI datasets, the presented algorithms are compared with three state-of-the-art meta-heuristics included Binary Genetic Algorithm (BGA), Binary Particle swarm optimization (BPSO) and Binary Atom Search Optimization (BASO). Due to its superior performance in solving feature selection problems, the proposed algorithm is optimal.

Keywords:

Feature selection , Golden Ratio Optimization , Accuracy , Meta-heuristic