Uncertainty-Aware Decomposed Hybrid Networks

2025-03-24

AI Summary

Existing few-shot image recognition methods suffer from poor robustness and limited interpretability, particularly under data-scarce conditions. Method: This paper proposes an uncertainty-aware decomposed hybrid network that synergistically integrates the adaptive representation capability of deep neural networks with domain-knowledge-driven quasi-invariant operators. It introduces a task-driven operator decomposition architecture and a confidence scoring mechanism tailored for multiple operators, explicitly modeling feature reliability and noise susceptibility. Contribution/Results: The approach significantly enhances model transparency and low-data generalization capability. On traffic sign detection, it achieves state-of-the-art performance under both semi-supervised and unsupervised settings, demonstrating its effectiveness and practicality in data-constrained scenarios.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Interpretability, Explainability, and Transparency
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

The robustness of image recognition algorithms remains a critical challenge, as current models often depend on large quantities of labeled data. In this paper, we propose a hybrid approach that combines the adaptability of neural networks with the interpretability, transparency, and robustness of domain-specific quasi-invariant operators. Our method decomposes the recognition into multiple task-specific operators that focus on different characteristics, supported by a novel confidence measurement tailored to these operators. This measurement enables the network to prioritize reliable features and accounts for noise. We argue that our design enhances transparency and robustness, leading to improved performance, particularly in low-data regimes. Experimental results in traffic sign detection highlight the effectiveness of the proposed method, especially in semi-supervised and unsupervised scenarios, underscoring its potential for data-constrained applications.

Problem

• Enhancing image recognition robustness with hybrid networks
• Reducing dependency on large labeled datasets
• Improving performance in low-data and unsupervised scenarios

Innovation

• Hybrid neural networks with quasi-invariant operators
• Task-specific decomposed operators with confidence measurement
• Enhanced robustness in low-data traffic sign detection