Single-Agent Finite Impulse Response Optimizer for Numerical Optimization Problems

This paper introduces a new single-agent metaheuristic optimization algorithm, named single-agent finite impulse response optimizer (SAFIRO). This proposed algorithm is inspired by the estimation ability of the ultimate iterative unbiased finite impulse response (UFIR) filter. The UFIR filter is one of the variants of the finite impulse response (FIR) filter, whereby in state space models, the FIR filter can be used as an option other than the Kalman filter (KF) for state estimation. Unlike the KF, the UFIR filter does not require any noise covariance, error covariance, and initial condition to calculate the state estimate. The UFIR filter also provides an iterative Kalman-like form to improve the estimation process. In the SAFIRO algorithm, the agent works as an individual UFIR to find an optimal or a near-optimal solution, where the agent needs to perform two main tasks; measurement and estimation. The performance of the SAFIRO algorithm is evaluated using the CEC 2014 Benchmark Test Suite for single-objective optimization and statistically compared with the several well-known metaheuristic optimization algorithms, such as Particle Swarm Optimization algorithm, Genetic Algorithm, and Grey Wolf Optimization algorithm. The experimental results show that the proposed SAFIRO algorithm is able to converge to the optimal and the near-optimal solutions, and significantly outperform all the aforementioned state-of-the-art metaheuristic algorithms.