Supervised Learning of Multievent Transition Matrices in Fuzzy Discrete-Event Systems

Area of Science:

• Computer Science
• Artificial Intelligence
• Control Systems

Background:

• Fuzzy discrete-event systems (FDES) offer a framework for modeling complex systems with inherent uncertainty.
• Explainability is crucial in practical applications, particularly in biomedical fields, necessitating interpretable system models.

Purpose of the Study:

• To develop and investigate a supervised learning algorithm for fuzzy discrete-event systems (FDES).
• To enable the learning of multievent transition matrices and event sequences within FDES models.
• To facilitate the creation of explainable models for complex systems using available data.

Main Methods:

• Derivation of a supervised learning algorithm tailored for FDES.
• Application of the algorithm to learn transition matrices and event sequences.
• Utilizing MATLAB for simulations to validate the algorithm's performance.

Main Results:

• The proposed algorithm effectively performs supervised learning for FDES.
• The algorithm successfully learns events and event sequences from data.
• Simulations confirm the algorithm's effectiveness in building explainable models.

Conclusions:

• The developed learning algorithm enhances the explainability of FDES models.
• System developers can leverage this algorithm to model complex systems from data.
• The approach holds significant potential for applications requiring interpretable models, such as in biomedicine.