![[2602.19253] Alternating Bi-Objective Optimization for Explainable Neuro-Fuzzy Systems](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19253] Alternating Bi-Objective Optimization for Explainable Neuro-Fuzzy Systems
Summary
This article presents X-ANFIS, a novel optimization scheme for explainable neuro-fuzzy systems that balances accuracy and explainability through alternating bi-objective gradient-based optimization.
Why It Matters
The research addresses the critical trade-off between accuracy and explainability in AI systems, particularly in fuzzy logic applications. By proposing a new method that improves upon existing multi-objective optimization techniques, this work contributes to the development of more interpretable AI models, which is essential for trust and transparency in AI applications.
Key Takeaways
- X-ANFIS improves the explainability of neuro-fuzzy systems while maintaining predictive accuracy.
- The method utilizes Cauchy membership functions for stable training.
- It introduces a differentiable explainability objective, enhancing the optimization process.
- The approach outperforms traditional multi-objective optimization in recovering non-convex Pareto regions.
- Validated across nine UCI regression datasets, demonstrating robust performance.
Computer Science > Machine Learning arXiv:2602.19253 (cs) [Submitted on 22 Feb 2026] Title:Alternating Bi-Objective Optimization for Explainable Neuro-Fuzzy Systems Authors:Qusai Khaled, Uzay Kaymak, Laura Genga View a PDF of the paper titled Alternating Bi-Objective Optimization for Explainable Neuro-Fuzzy Systems, by Qusai Khaled and 2 other authors View PDF HTML (experimental) Abstract:Fuzzy systems show strong potential in explainable AI due to their rule-based architecture and linguistic variables. Existing approaches navigate the accuracy-explainability trade-off either through evolutionary multi-objective optimization (MOO), which is computationally expensive, or gradient-based scalarization, which cannot recover non-convex Pareto regions. We propose X-ANFIS, an alternating bi-objective gradient-based optimization scheme for explainable adaptive neuro-fuzzy inference systems. Cauchy membership functions are used for stable training under semantically controlled initializations, and a differentiable explainability objective is introduced and decoupled from the performance objective through alternating gradient passes. Validated in approximately 5,000 experiments on nine UCI regression datasets, X-ANFIS consistently achieves target distinguishability while maintaining competitive predictive accuracy, recovering solutions beyond the convex hull of the MOO Pareto front. Comments: Subjects: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE) MSC classes: ...
