Improving fuzzy cognitive maps learning through memetic particle swarm optimization

Abstract

Fuzzy cognitive maps constitute a neuro-fuzzy modeling methodology that can simulate complex systems accurately. Although their configuration is defined by experts, learning schemes based on evolutionary and swarm intelligence algorithms have been employed for improving their efficiency and effectiveness. This paper comprises an extensive study of the recently proposed swarm intelligence memetic algorithm that combines particle swarm optimization with both deterministic and stochastic local search schemes, for fuzzy cognitive maps learning tasks. Also, a new technique for the adaptation of the memetic schemes, with respect to the available number of function evaluations per application of the local search, is proposed. The memetic learning schemes are applied on four real-life problems and compared with established learning methods based on the standard particle swarm optimization, differential evolution, and genetic algorithms, justifying their superiority.

A review on methods and software for fuzzy cognitive maps

“…A memetic approach that combines PSO with both deterministic and stochastic local search schemes for solving the above problem was investigated in [110] [111]. The memetic schemes were applied to real-life problems and compared with wellknown continuous optimizers, thus justifying their superiority.…”

Integrating data, expert opinion and fuzzy logic in the development of an index of wetland condition

“…Using a stochastic global optimisation routine based on a particle swarm algorithm (e.g. Petalas et al 2009), estimates of the best-fitting edge weights for the FCM were determined. The problem of over-training bias, a phenomenon in which the model predicts the observed data at hand well but predicts new data poorly, was reduced using a cross-validation procedure.…”

ICLA imperialist competitive learning algorithm for fuzzy cognitive map: Application to water demand forecasting

“…They are often used to solve optimization problems. Several population-based algorithms, such as GS [11], (GA) [9], RCGA [30,31,33], Chaotic Simulated Annealing [2], TS [3], game-based learning [14], Particle Swarm Optimization (PSO) [25], Memetic PSO (MPSO) [27], ACO [7], Extended Great Deluge algorithm [6], Big Bang-Big Crunch [38], Artificial Bee Colony [37], cellular automata [16], immune algorithm [12], and supervised gradient-based algorithm [36] have been proposed for training FCMs. The first evolutionary learning algorithm was developed in 2001 by Koulouriotis et al [11].…”

“…Previous solutions were developed by implementing several evolutionary or population based algorithms [19,29,33]. Common evolutionary optimization approaches used for FCM learning are genetic strategies (GS) [11], genetic algorithms (GA) [9], real coded generic algorithm -RCGA [30,31,33], Swarm Intelligence [25], Chaotic Simulated Annealing [2], Tabu search [3], game-based learning [14], Particle Swarm Optimization (PSO) [25], Memetic PSO (MPSO) [27], Ant Colony Optimization (ACO) [7], Extended Great Deluge algorithm [6], Big Bang-Big Crunch [38], Artificial Bee Colony (ABC) [37].…”