Quantum-Membership-Function-Based Adaptive Neural Fuzzy Inference System
The well-known ANFIS (adaptive-network-based fuzzy inference system) has demonstrated good performance and applicability. This chapter presents a quantum version of ANFIS model, namely, qANFIS, whose network structure is similar to ANFIS but it is with quantum membership functions (qMF). A hybrid learning procedure is proposed to update the parameters. First, we develop a genetic algorithm with trimming operator to determine the number of quantum levels in qMF. Second, the least squares estimate method is applied to update the consequent parameters. Finally, we introduce two methods to update the qMF parameters, the gradient descent and the quantum-inspired particle swarm optimization methods, and one of them is selected for training purpose. The simulation results of path planning had demonstrated that the proposed method could reach satisfactory performance and smaller error measure as compared with ANFIS.
KeywordsANFIS Genetic algorithm Neuro-fuzzy system Path planning Quantum function Quantum particle swarm optimization
This work was supported by the National Science Council of Taiwan (R.O.C.) under grant no. NSC 100-2410-H-364-002.
- 1.Chai Y, Jia L, Zhang Z (2009) Mamdani model based adaptive neural fuzzy inference system and its application. Int J Comput Intell 5:22–29Google Scholar
- 3.Chiang C-H (2008) A symbolic controller based intelligent control system with quantum particle swarm optimization based hybrid genetic algorithm. In: Congress on evolutionary computation. IEEE Press, New York, pp 1356–1363Google Scholar
- 7.Lu KZ, Wang RC (2008) Application of PSO and QPSO algorithm to estimate parameters from kinetic model of glutamic acid batch fermentation. In: 7th world congress on intelligent control and automation. IEEE Press, New York, pp 8968–8971Google Scholar