Fragility-aware Classification for Understanding Risk and Improving Generalization

2025-02-18

AI Summary

Conventional classification models, when deployed in high-stakes domains such as medical diagnosis, rely on aggregate metrics like accuracy and neglect the confidence of erroneous predictions—leading to hazardous overconfident misclassifications. Method: We propose the Fragility Index (FI), the first quantitative measure targeting tail-risk exposure, which quantifies a model’s sensitivity to high-confidence errors. Building upon FI, we introduce the Robust Satisficing (RS) optimization framework—a unified, uncertainty-aware training paradigm compatible with cross-entropy, hinge-type, and Lipschitz-constrained losses. Contribution/Results: Evaluated on synthetic benchmarks and real-world medical diagnostic tasks, RS significantly enhances model robustness and out-of-distribution generalization while effectively suppressing overconfident mispredictions. The approach delivers reliable, cost-sensitive, and safety-critical classification guarantees without compromising standard performance.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Safety and Robustness
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Classification models play a critical role in data-driven decision-making applications such as medical diagnosis, user profiling, recommendation systems, and default detection. Traditional performance metrics, such as accuracy, focus on overall error rates but fail to account for the confidence of incorrect predictions, thereby overlooking the risk of confident misjudgments. This risk is particularly significant in cost-sensitive and safety-critical domains like medical diagnosis and autonomous driving, where overconfident false predictions may cause severe consequences. To address this issue, we introduce the Fragility Index (FI), a novel metric that evaluates classification performance from a risk-averse perspective by explicitly capturing the tail risk of confident misjudgments. To enhance generalizability, we define FI within the robust satisficing (RS) framework, incorporating data uncertainty. We further develop a model training approach that optimizes FI while maintaining tractability for common loss functions. Specifically, we derive exact reformulations for cross-entropy loss, hinge-type loss, and Lipschitz loss, and extend the approach to deep learning models. Through synthetic experiments and real-world medical diagnosis tasks, we demonstrate that FI effectively identifies misjudgment risk and FI-based training improves model robustness and generalizability. Finally, we extend our framework to deep neural network training, further validating its effectiveness in enhancing deep learning models.

Problem

• Develops Fragility Index for risk evaluation.
• Enhances model robustness and generalizability.
• Applies risk-averse metric in deep learning.

Innovation

• Introduces Fragility Index (FI)
• Optimizes FI with robust satisficing
• Extends FI to deep learning models