Optimizing Support Vector Regression with Swarm Intelligence for Estimating the Concrete Compression Strength

Abstract

Estimating the compression strength of concrete is a complex problem which has been studied by various researchers. Support Vector Regression (SVR) is a technique that has been shown to be adequate for estimation through input examples. In this paper, we assess three swarm algorithms, Fish School Search (FSS), Artificial Bee Colony (ABC) and Particle Swarm Optimization (PSO) aiming to optimize the SVR parameter. The results show that both all swarm-based algorithms far outperformed the original SVR in the concrete compression strength estimation task and the FSS and ABC obtained better results than PSO for this particular problem.