报告人简介
Dr. Hao Ying is a Professor in the Department of Electrical and Computer Engineering at Wayne State University, USA. He received the IEEE CIS Fuzzy Systems Pioneer Award in 2023—the highest honor bestowed by the IEEE Computational Intelligence Society. Prof. Ying currently serves as an Associate Editor or Editorial Board Member for 20 international journals, including the IEEE Transactions on Fuzzy Systems and the IEEE Transactions on Systems, Man, and Cybernetics: Systems. He is an IEEE Fellow.
报告摘要
To effectively represent deterministic uncertainties and vagueness as well as human subjective observation and judgment encountered in many real-world problems especially those in medicine, we originated a theory of fuzzy discrete event systems (DES) in 2001. The theory is unique in that it is capable of modeling a class of event-driven systems as fuzzy automata with states and event-invoked state transitions being ambiguous. We introduced fuzzy states and fuzzy event transition and generalized conventional crisp DES to fuzzy DES.
Other researchers have extended the FDES framework afterward.They studied a variety of fundamental FDES-related issues. These include supervisory control, observability, decentralized control, diagnosability, online control, state-based control, state-feedback control, prognosis, predictability, opacity, and detectabilities.
For a concrete example of an FDES and its operation, refer to the illustrative numerical FDES example included in the Fuzzy Logic Toolbox of MATLAB, version R2024a
In this presentation, we show our recent work on how a FDES can learn the event transition matrix of its fuzzy automaton, which is the core of the FDES that is difficult to build manually, when neither pre-event nor post-event state is known. The only information assumes to be available is values of the variables that are only known to be vaguely related to the pre- and post-event states. We link the variables to the states through Gaussian fuzzy sets owing to the conceptual relationship between a fuzzy set and a fuzzy state. Stochastic-gradient-descent-based algorithms are derived for the model to iteratively learn online the event transition matrix and all the parameters of the fuzzy sets simultaneously. This leads to the first FDES model that is capable of learning solely based on sensor data without relying on any subjective input from humans, removing a significant bottleneck for real-world applications.
We will also present a new class of FDES called the stochastic fuzzy discrete event systems (SFDES) that we introduced lately. A SFDES is comprised of multiple fuzzy automata that occur randomly one at time with different occurrence probabilities. Leveraging the SFDES framework, we extended conventional Markov chains to fuzzy Markov chainscapable of handling both fuzzy states and fuzzy events, which is impossible for the conventional Markov chains. Crucially, the fuzzy Markov chain fully preserves the stochastic characteristics defined by the transition probability matrix of the binary Markov chain, ensuring identical stochastic behaviors. The former encompasses the latter as a special case.
