A novel self-organizing neural fuzzy network for automatic generation of fuzzy inference systems

Authors:
Meng Joo Er;Rishikesh Parthasarathi
Affiliations:
Intelligent Systems Center (IntelliSys), Nanyang Technological University, Singapore;Intelligent Systems Center (IntelliSys), Nanyang Technological University, Singapore
Venue:
ISNN'05 Proceedings of the Second international conference on Advances in Neural Networks - Volume Part I
Year:
2005

Citing 2
Cited 0

Introduction to Reinforcement Learning

Introduction to Reinforcement Learning
Learning to Predict by the Methods of Temporal Differences

Machine Learning

Quantified Score

Hi-index	0.01

Visualization

Abstract

This paper presents Fuzzy Multi-Agent Structure Learning (FMASL), a neural fuzzy network for unsupervised clustering and automatic structure generation of Fuzzy Inference Systems (FISs). The FMASL clustering identifies crisp clusters in an unlabeled input data and represents them by an agent, using competitive agent learning. In generating a FIS, the FMASL is used to identify the optimum number of agents (rules) of the FIS. The best action (consequent) for each agent is automatically selected using an enhanced version of Actor-Critic learning (ACL). The structure of the FIS is dynamically changed based only on experiences and no expert's knowledge is required. This is a significant feature of our approach because constructing a FIS manually for a complex task is very difficult, if not impossible. The performance of the algorithm is elucidated using the cart-pole balancing problem.