Fully Explainable Classification Models Using Hyperblocks

2025-06-08

AI Summary

Classification models often face a fundamental trade-off among interpretability, accuracy, training efficiency, and model complexity. Method: This paper proposes Hyperblocks—a fully interpretable classification framework that defines explicit, semantically transparent decision regions via attribute upper and lower bounds. It introduces a hyperblock simplification algorithm (incorporating redundant-attribute/block elimination and disjunctive-unit construction) and an interpretable k-NN fallback mechanism, ensuring zero coverage gaps while achieving lightweight design and robustness. Contribution/Results: Evaluated on WBC (9D) and MNIST (784D), the framework matches black-box models in accuracy while drastically reducing parameter count. It demonstrates scalable synergy between high-fidelity interpretability and predictive performance in high-dimensional settings, enabling domain experts—without ML expertise—to intuitively understand, validate, and refine decision logic.

Technology Category

• Machine Learning: Transparent, Interpretable, Explainable ML
• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Building on existing work with Hyperblocks, which classify data using minimum and maximum bounds for each attribute, we focus on enhancing interpretability, decreasing training time, and reducing model complexity without sacrificing accuracy. This system allows subject matter experts (SMEs) to directly inspect and understand the model's decision logic without requiring extensive machine learning expertise. To reduce Hyperblock complexity while retaining performance, we introduce a suite of algorithms for Hyperblock simplification. These include removing redundant attributes, removing redundant blocks through overlap analysis, and creating disjunctive units. These methods eliminate unnecessary parameters, dramatically reducing model size without harming classification power. We increase robustness by introducing an interpretable fallback mechanism using k-Nearest Neighbor (k-NN) classifiers for points not covered by any block, ensuring complete data coverage while preserving model transparency. Our results demonstrate that interpretable models can scale to high-dimensional, large-volume datasets while maintaining competitive accuracy. On benchmark datasets such as WBC (9-D), we achieve strong predictive performance with significantly reduced complexity. On MNIST (784-D), our method continues to improve through tuning and simplification, showing promise as a transparent alternative to black-box models in domains where trust, clarity, and control are crucial.

Problem

• Enhancing interpretability of Hyperblock classification models without accuracy loss
• Reducing Hyperblock complexity via attribute and block simplification algorithms
• Ensuring complete data coverage with interpretable k-NN fallback mechanism

Innovation

• Hyperblocks with simplified algorithms for interpretability
• Interpretable fallback using k-NN for uncovered points
• Reduced model complexity without losing accuracy