A Human-in-the-Loop Framework for Interactive and Explainable Data-Driven Modeling

作者：Sungwook HONG, Wen Yu, Tianyou Chai · 发表于：IEEE Transactions on Emerging Topics in Computational Intelligence · 年份：2025 · DOI：10.1109/tetci.2025.3637807 · 被引用次数：1 · 研究领域：Explainable Artificial Intelligence (XAI)、Machine Learning in Healthcare、Multimodal Machine Learning Applications

Data-driven models are widely adopted, but their opaque nature often hinders understanding and trust. This paper introduces a novel human-in-the-loop framework for interactive and explainable data-driven modeling. Central to our approach is the integration of fuzzy neural networks (FNNs) with evolutionary optimization, guided by continuous human feedback. Unlike traditional methods, our framework enables users to actively participate, influencing the FNN’s decision-making in real-time. This feedback directly informs the model’s structural and parameter updates, resulting in models that are both dynamically adaptable and inherently transparent through their fuzzy logic. Experimental results demonstrate this framework significantly enhances model performance while fostering trust and comprehension by providing clear reasoning processes. This work advances human-centric data modeling, improving explainability and interactivity for complex data relationships.