The Self-Organizing Relationship (SOR) network employing fuzzy inference based heuristic evaluation

Area of Science:

• Computational intelligence
• Control systems engineering
• Machine learning

Background:

• Human skill acquisition relies on evaluating numerous experiences using criteria derived from knowledge or advice.
• The Self-Organizing Relationship (SOR) network computationally emulates this learning process.
• Previous SOR network applications used mathematical expressions for evaluation criteria, which become challenging for complex systems.

Purpose of the Study:

• To enhance Self-Organizing Relationship (SOR) network controller design by incorporating heuristic evaluation criteria.
• To address the limitations of mathematical expressions in defining evaluation criteria for complex systems.
• To leverage fuzzy inference for realizing human-like heuristic evaluation in computational models.

Main Methods:

• The study employs fuzzy inference to represent heuristic evaluation criteria within the Self-Organizing Relationship (SOR) network framework.
• This method allows for the computational emulation of human-like judgment in complex control scenarios.
• The integration of fuzzy logic aims to simplify the definition of success/failure criteria compared to purely mathematical approaches.

Main Results:

• Fuzzy inference successfully enables the use of heuristic expressions for evaluation criteria in SOR networks.
• This approach offers a more practical method for designing controllers for complex systems where mathematical criteria are difficult to formulate.
• The findings suggest improved adaptability and performance in SOR networks when using heuristic evaluation.

Conclusions:

• Employing fuzzy inference within SOR networks provides a viable alternative to mathematical expressions for defining evaluation criteria.
• This integration facilitates the design of controllers for complex systems by mimicking human heuristic evaluation.
• The study highlights the potential of fuzzy logic in advancing computational learning and control system design.