A continuous variable neighbourhood search based on specialised EAs
Carlos García‐Martínez, Manuel Lozano · 2009
Variable Neighbourhood Search is a metaheuristic combining three components: generation, improvement, and shaking components. In this paper, we design a continuous Variable Neighbourhood Search algorithm based on three specialised Evolutionary Algorithms, which play the role of each aforementioned component: 1) an EA specialised in generating a good starting point as generation component, 2) an EA specialised in exploiting local information as improvement component, 3) and another EA specialised in providing local diversity as shaking component. Experiments are carried out on the noiseless Black-Box Optimisation Benchmark 2009 testbed.