A Temporal Type-2 Fuzzy System for Time-dependent Explainable Artificial Intelligence

Fuzzy Logic offers a practical path to explainable AI: The paper demonstrates that Fuzzy Logic Systems can provide transparent AI models that non-technical audiences can understand and analyze, making it a viable alternative to "black box" AI systems that are difficult to interpret.

Time matters for real-world AI applications: Most real-world processes change over time, yet traditional explainable AI systems don't account for this. The new Temporal Type-2 Fuzzy approach addresses this gap by incorporating how measurements change and recur over time, which is critical for applications like healthcare monitoring or financial forecasting.

Significant performance improvements with temporal awareness: The temporal approach achieved a mean recall of 95.40% compared to 87.04% for standard explainable systems—a meaningful improvement that suggests accounting for time-dependent patterns makes AI systems both more accurate and more interpretable.

Better performance than many non-explainable systems: The temporal explainable AI system outperformed most conventional (non-explainable) AI systems by 3.95% to 19.04%, suggesting that explainability doesn't require sacrificing accuracy—a key concern for organizations considering responsible AI adoption.

Addresses uncertainty in real-world environments: By modeling both uncertainty and temporal patterns, this approach is better suited for complex real-world applications where decisions must account for changing conditions and incomplete information.