scientific article; zbMATH DE number 1349973
Introductory exposition (textbooks, tutorial papers, etc.) pertaining to statistics (62-01) Research exposition (monographs, survey articles) pertaining to statistics (62-02) Time series, auto-correlation, regression, etc. in statistics (GARCH) (62M10) Inference from stochastic processes and prediction (62M20) Learning and adaptive systems in artificial intelligence (68T05) Theory of languages and software systems (knowledge-based systems, expert systems, etc.) for artificial intelligence (68T35) Fuzzy control/observation systems (93C42) Fuzzy sets and logic (in connection with information, communication, or circuits theory) (94D05)
The book is devoted to the study of combined techniques to be applied to prediction models and based on Takagi-Sugeno fuzzy systems (TSFSs), with special emphasis on parametric and structural control problems. Chapter 1 surveys the most important statistical models of prediction. Chapter 2 (Clustering and Fuzzy Systems) describes TSFSs whose knowledge base rules may encode expertise explicitly, using an a priori knowledge on the problem, or implicitly, starting from a set of learning data. One main difference between TSFSs and classical knowledge-based systems is the multi-valued logic that TSFSs are relying on. Using fuzzy subsets to encode the various states of a linguistic variable, TSFSs may activate simultaneously several rules that have different (truth) degrees. Another specific feature of TSFSs is that learning within these systems is done in two basic manners: local learning, by competition among the artificial experts, and global learning, by their cooperation. Chapter 3 (Parametrical and Structural Control of TSFSs) proposes a generalized TSFS using an (automatic structural) fitting method, of decreasing type, based on a study of the rule sensitivity. A new parameter is introduced for detecting and removing the system redundancy. This technique is successfully applied to prediction based on two highly nonlinear chaotique series: Mackey-Glass series and Laser series. Chapter 4 (Aggregation of Prediction Models) applies the proposed structural fitting to the method of combined predictions. Several techniques are discussed: the method of optimal weights, linear regression methods, and Bayesian methods. The final Chapter 5 (Takagi-Sugeno Models for the Aggregation of Prediction Models) contains tests and simulations for the proposed methods, concentrating on their performance, complexity, and readability of the generalized TSFSs. Some interesting conclusions are drawn: (1) the learning type of internal parameters for the fuzzy system is essential; (2) the global training of fuzzy systems allows to successfully combine various prediction methods. Three useful appendices are enclosed, and devoted to: (A) Conditional (mathematical) expectation; (B) Estimation and regression; (C) Resolution of linear least squares.
- Fuzzy time series and its models
- Computational intelligence in time series forecasting. Theory and engineering applications
- Regression prediction for fuzzy time series
- Mamdani-Larsen fuzzy system based on expectation maximization algorithm and its applications to time series prediction
- scientific article; zbMATH DE number 1487537 (Why is no real title available?)
- Generalized dynamical fuzzy model for identification and prediction
- scientific article; zbMATH DE number 7112985 (Why is no real title available?)
- Local identification of prototypes for genetic learning of accurate TSK fuzzy rule-based systems
- A method of fast learning fuzzy model for nonlinear systems
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4265382)