A Temporal Type-2 Fuzzy System for Time-Dependent Explainable Artificial Intelligence
Mehrin Kiani, Javier Andreu-Pérez, Hani Hagras
Abstract
Explainable artificial intelligence (XAI) focuses on transparent AI models and decisions, which are easy to understand, analyze, and augment by a nontechnical audience. Fuzzy logic systems (FLS)-based XAI provides an explainable framework while also modeling uncertainties in real-world environments. However, most real-life processes are not characterized by high uncertainty alone; they are also inherently time dependent, i.e., the processes are time variant. In this work, we present a novel temporal type-2 FLS-based approach for time-dependent XAI (TXAI) systems, which can account for the likelihood of a sample occurrence in the time domain by its the frequency. In the proposed temporal type-2 fuzzy sets (TT2FSs), a 4-D time-dependent membership function integrates the universe of discourse, its membership, and its frequency of occurrence across time. The TXAI system manifested better classification prowess in cross-validation tests, with a mean recall of 95.40% than a standard XAI system (based on nontemporal general type-2 fuzzy sets) that had a mean recall of 87.04%. TXAI also performed significantly better than most nonexplainable AI systems, with between 3.95% and 19.04% improvement gain in mean recall. In addition, TXAI can also outline the most likely time-dependent trajectories using the frequency and time dimensions embedded in the TXAI model; viz. given a rule at a determined time interval, what will be the next most likely rule at a subsequent time interval. In this regard, the proposed TXAI system can have profound implications for delineating the evolution of real-life time-dependent processes, such as behavioral or biological processes.