A novel optimization booster algorithm

• Published in: Computers & industrial engineering 2019-10, Vol.136, p.591-613
• Main Authors: Pakzad-Moghaddam, S.H., Mina, Hassan, Mostafazadeh, Parisa
• Format: Article
• Language: English
• Subjects: Artificial intelligence; Computer science; Evolutionary computation; Exchange market; Mathematical optimization
• ISSN: 0360-8352; 1879-0550
• DOI: 10.1016/j.cie.2019.07.046

[Display omitted] •Introducing a new meta-heuristic called Optimization Booster Algorithm (i.e. OBA).•OBA is inspired by human intelligent behavior in exchange markets.•Boosted algorithms often provided better solutions, while consuming less time.•OBA has improved feasibility, optimality and efficiency of the final results. In this paper, a novel meta-heuristic method called the Optimization Booster Algorithm (OBA) is presented. It incorporates existing optimization methods with human-inspired intelligence, which applies particularly while conducting business in exchange markets. A key objective in exchange markets is to increase wealth over time, which is typically the same objective when performing optimizations. Moreover, optimization is about finding a way to increase the fitness value of a system, by spending adequate computation time. The OBA is founded on the core idea that a key reason behind the rapid evolution of human societies compared to the tortoise-like natural evolution is the forward-looking approach. In exchange markets, analysts have learned to make decisions based on forecasted prices, rather than the current prices; and this is an illustrative example of the application of such forward-looking approaches, which form the essence of the OBA. Following extensive numerical experiments, and applications of fourteen well-known heuristic and meta-heuristic methods to solve seventy-one non-linear unconstrained and constrained, single-objective and multi-objective benchmarks, before and after receiving a boost, the OBA performance is investigated. It has proven — in both theory and practice — to quite significantly improve existing optimization methods. In most cases, boosting resulted in much better quality of outputs, while requiring less computation time.