CAFD: Concept-Aware DNN Fault Detection using VLMs

2026-05-19

AI Summary

This work addresses the high computational cost and practical limitations of existing deep neural network fault detection methods when fusing multi-source information. To overcome this challenge, the authors propose CAFD, a learning-based, efficient fault detection framework that introduces a novel feature termed Concept Failure Ratio (CFR). CAFD leverages vision-language models to extract semantic concepts from images that correlate with model failures and synergistically integrates these with model output signals and distance-based features for comprehensive multi-source fusion. Experimental results on three benchmarks, including ImageNet, demonstrate that CAFD consistently outperforms five state-of-the-art baselines, achieving an average improvement of 18.3% in fault detection rate across various selection budgets.

Abstract

Fault detection for Deep Neural Networks (DNNs) has received increasing attention in recent years. While more advanced hybrid approaches have been proposed to combine multiple sources of information and outperform earlier techniques, they often incur substantial computational overhead, limiting scalability and practicality in real-world settings. In this paper, we introduce Concept-Aware Fault Detection (CAFD), a learning-based approach that achieves superior fault detection performance by effectively integrating multiple information sources while maintaining practical efficiency. Specifically, CAFD is trained using a carefully selected set of informative features, including model-based signals derived from the DNN's outputs, distance-based features, and a novel concept-based feature, called Concept Failure Ratio (CFR). CFR leverages Vision-Language Models (VLMs) to extract textual concepts from images and quantify the likelihood that their presence is associated with DNN failures. By incorporating this feature, CAFD benefits from complementary semantic information, enabling more effective fault detection. Our results demonstrate that CFR serves as an effective indicator for DNN fault detection. We conduct an extensive empirical evaluation of CAFD, comparing it against five state-of-the-art baselines across three subject DNN models and datasets, including ImageNet. Across a wide range of constrained selection budgets, CAFD consistently outperforms all baselines in Fault Detection Rate (FDR), achieving average FDR improvements of 18.3% across all investigated subjects and budget sizes.

Problem

• DNN fault detection
• computational overhead
• scalability
• practicality
• hybrid approaches

Innovation

• Concept Failure Ratio
• Vision-Language Models
• Fault Detection
• Deep Neural Networks
• Semantic Features