Modified Binary Artificial Bee Colony Algorithm with New Exploration-Exploitation Strategies

In this study, a novel binary optimization algorithm which addresses the exploitation problem of Artificial Bee Colony (ABC) algorithm is presented and named as Scoutless Rule-driven binary Artificial Bee Colony (SRABC). The algorithm exhibits competitive performance through comparison to recent binary ABC variants applied in feature selection problems. Similarly, an experimental verification is conducted for actual case scenario which showed that the algorithm is efficient in increasing the prediction accuracy and discovered the highly important predictors for on-time graduation. The better performance of SRABC is attributed to the new strategies applied in the employed and onlooker bee phases and the turning back of food sources in initialization phase.