Hybrid FCM learning algorithm based on particle swarm optimization and gradient descent algorithm
- 13 December 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 801-806
Abstract
The learning algorithm of fuzzy cognitive map(FCM) is mainly based on expert intervention or effective historical data to learn the weight value in weight matrix W. With the increase of the number of nodes, the number of weights need to be learned will also increase square, the performance of the learning algorithm will decline sharply. Therefore, it is an important task to design a high-performance learning algorithm which depends on expert knowledge as little as possible. A single learning algorithm has some limitations, such as easy local optimization, slow learning rate and so on. Among them, particle swarm optimization algorithm has strong global optimization ability, which can quickly optimize the weight matrix to a certain accuracy, but the follow-up optimization time is too long. For the gradient descent algorithm, when the initial value is chosen properly, its can approach the global optimal solution quickly. In this paper, we design particle swarm optimization (PSO) and gradient descent algorithm (GDA) to learn FCM respectively. According to the characteristics of the two algorithms, a hybrid algorithm(HA) is proposed: first, the PSO algorithm is used to optimize weight matrix to a certain accuracy, then the gradient descent method is added, which makes up for the disadvantage that the error of single PSO algorithm decreases slowly in the later stage. Experimental results show that the hybrid algorithm proposed in this paper has a significant improvement in both accuracy and speed.Keywords
This publication has 15 references indexed in Scilit:
- A Genetically Optimized Method for Weight Updating in Fuzzy Cognitive Maps2018
- Uncertainty Propagation in Fuzzy Grey Cognitive Maps With Hebbian-Like Learning AlgorithmsIEEE Transactions on Cybernetics, 2017
- Fuzzy cognitive maps for decision making in triage of non-critical elderly patients2017
- A Mutual Information-Based Two-Phase Memetic Algorithm for Large-Scale Fuzzy Cognitive Map LearningIEEE Transactions on Fuzzy Systems, 2017
- Interval-valued fuzzy cognitive maps for supporting business decisions2016
- Learning Fuzzy Cognitive Maps using Structure Optimization Genetic Algorithm2015
- Learning fuzzy cognitive maps using decomposed parallel ant colony algorithm and gradient descent2015
- Learning of Fuzzy Cognitive Maps With Varying Densities Using A Multiobjective Evolutionary AlgorithmIEEE Transactions on Fuzzy Systems, 2015
- Fuzzy Cognitive Maps: a model for intelligent supervisory control systemsComputers in Industry, 1999
- Fuzzy cognitive mapsInternational Journal of Man-Machine Studies, 1986