Lookback-Guess-Next Optimizer: Feedback-Guided Random Search Technique with Biased Mapping for Solving Unconstrained Optimization Problems

Abstract

Finding global optima for functions is a very important problem. Although a large number of methods have been proposed for solving this problem, more effective and efficient methods are greatly required. This paper proposes an innovative method that combines different effective techniques for speeding up the convergence to the solution and greatly improving its precision. In particular, the method uses feedback-guided random search technique to identify the promising regions of the domains and uses the biased mapping technique to focus the search on these promising regions, without ignoring the other regions of the domains. Therefore, at any point of time, the domain of each variable is entirely covered with much more emphasis on the promising regions. Experiments with our prototype implementation showed that our method is efficient, effective, and outperformed the state-of-art techniques.