Towards the Formalization of a Trustworthy AI for Mining Interpretable Models explOiting Sophisticated Algorithms

2025-10-23

AI Summary

Balancing interpretability, predictive performance, and ethical attributes—causality, fairness, and privacy—remains a fundamental challenge in trustworthy AI. Method: This work proposes MIMOSA, a unified framework that formally characterizes three canonical paradigms of interpretable modeling and defines a multidimensional explanation structure. It intrinsically integrates causal inference, fairness-aware optimization, and differential privacy mechanisms directly into the modeling pipeline, supporting heterogeneous modalities including tabular, time-series, image, and text data. Contribution/Results: MIMOSA establishes the first verifiable evaluation framework jointly incorporating causality, fairness, and privacy, uncovering their inherent trade-offs. It provides rigorous theoretical foundations and practical algorithms enabling the co-optimization of high accuracy, model transparency, and ethical compliance. Empirically validated across diverse benchmarks, MIMOSA delivers a systematic methodology for deploying trustworthy AI in high-stakes decision-making scenarios.

Technology Category

• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Machine Learning: Ethics, Bias, and Fairness
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Security and Privacy: Data transparency and provenance
• Responsible Web: Algorithmic accountability and transparency on the web

Abstract

Interpretable-by-design models are crucial for fostering trust, accountability, and safe adoption of automated decision-making models in real-world applications. In this paper we formalize the ground for the MIMOSA (Mining Interpretable Models explOiting Sophisticated Algorithms) framework, a comprehensive methodology for generating predictive models that balance interpretability with performance while embedding key ethical properties. We formally define here the supervised learning setting across diverse decision-making tasks and data types, including tabular data, time series, images, text, transactions, and trajectories. We characterize three major families of interpretable models: feature importance, rule, and instance based models. For each family, we analyze their interpretability dimensions, reasoning mechanisms, and complexity. Beyond interpretability, we formalize three critical ethical properties, namely causality, fairness, and privacy, providing formal definitions, evaluation metrics, and verification procedures for each. We then examine the inherent trade-offs between these properties and discuss how privacy requirements, fairness constraints, and causal reasoning can be embedded within interpretable pipelines. By evaluating ethical measures during model generation, this framework establishes the theoretical foundations for developing AI systems that are not only accurate and interpretable but also fair, privacy-preserving, and causally aware, i.e., trustworthy.

Problem

• Formalizing interpretable AI models for trustworthy decision-making
• Balancing model interpretability with performance and ethics
• Embedding causality, fairness and privacy in interpretable pipelines

Innovation

• Formalizes interpretable model families and dimensions
• Integrates causality fairness privacy into model pipelines
• Evaluates ethical measures during model generation process