Readable Twins of Unreadable Models

2025-04-17

AI Summary

Deep learning models (DLMs) suffer from limited interpretability, hindering trust and deployment in high-stakes domains. Method: This paper introduces the “Readable Twin” framework—the first to adapt digital twin principles to eXplainable Deep Learning (XDL)—by constructing a lightweight, human-understandable proxy based on an Imprecise Information Flow Model (IIFM). We propose a systematic DLM-to-IIFM translation pipeline integrating model distillation, behavioral trajectory abstraction, and imprecise information flow modeling, balancing fidelity and interpretability. Contribution/Results: Evaluated on MNIST image classification, the generated IIFM twin significantly enhances decision transparency while preserving the original model’s classification logic with high consistency. Our core contribution is the formal definition and implementation of the first XDL-specific digital twin paradigm, enabling structured, traceable, and verifiable explanations for black-box models—thereby advancing both theoretical foundations and practical interpretability tools in XAI.

Technology Category

Machine Learning: Transparent, Interpretable, Explainable MLHumans and AI: Explainable AI (XAI) for Human UnderstandingComputer Vision: Interpretability, Explainability, and Transparency

Application Category

User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalizationSemantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactionsWeb Mining and Content Analysis: Large pretrained models with web data

Abstract

Creating responsible artificial intelligence (AI) systems is an important issue in contemporary research and development of works on AI. One of the characteristics of responsible AI systems is their explainability. In the paper, we are interested in explainable deep learning (XDL) systems. On the basis of the creation of digital twins of physical objects, we introduce the idea of creating readable twins (in the form of imprecise information flow models) for unreadable deep learning models. The complete procedure for switching from the deep learning model (DLM) to the imprecise information flow model (IIFM) is presented. The proposed approach is illustrated with an example of a deep learning classification model for image recognition of handwritten digits from the MNIST data set.

Problem

• Creating explainable deep learning systems for responsible AI
• Developing readable twins for unreadable deep learning models
• Transforming deep learning models into imprecise information flow models

Innovation

• Creating readable twins for unreadable models
• Using imprecise information flow models
• Transition from DLM to IIFM for explainability