Collaborative Learning of Scattering and Deep Features for SAR Target Recognition with Noisy Labels

2025-08-11

AI Summary

SAR target recognition suffers from scarce high-quality annotations and pervasive label noise, while existing image-based noisy-label learning methods fail to accommodate SAR’s non-intuitive scattering characteristics. To address this, we propose a scattering-feature–deep-semantic collaborative learning framework: (1) modeling physics-driven attribute scattering centers as dynamic graph structures; (2) employing inter-class Gaussian Mixture Models (GMMs) to discriminate clean from noisy samples; and (3) introducing a dual-branch semi-supervised co-training strategy with joint distribution alignment to enhance pseudo-label reliability. By tightly integrating SAR physical priors with deep representation learning, our method achieves state-of-the-art performance on the MSTAR dataset across diverse noise types and imaging conditions—outperforming existing denoising and automatic target recognition (ATR) approaches by significant margins.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Intelligent Robots: State Estimation
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

The acquisition of high-quality labeled synthetic aperture radar (SAR) data is challenging due to the demanding requirement for expert knowledge. Consequently, the presence of unreliable noisy labels is unavoidable, which results in performance degradation of SAR automatic target recognition (ATR). Existing research on learning with noisy labels mainly focuses on image data. However, the non-intuitive visual characteristics of SAR data are insufficient to achieve noise-robust learning. To address this problem, we propose collaborative learning of scattering and deep features (CLSDF) for SAR ATR with noisy labels. Specifically, a multi-model feature fusion framework is designed to integrate scattering and deep features. The attributed scattering centers (ASCs) are treated as dynamic graph structure data, and the extracted physical characteristics effectively enrich the representation of deep image features. Then, the samples with clean and noisy labels are divided by modeling the loss distribution with multiple class-wise Gaussian Mixture Models (GMMs). Afterward, the semi-supervised learning of two divergent branches is conducted based on the data divided by each other. Moreover, a joint distribution alignment strategy is introduced to enhance the reliability of co-guessed labels. Extensive experiments have been done on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset, and the results show that the proposed method can achieve state-of-the-art performance under different operating conditions with various label noises.

Problem

• SAR target recognition with noisy labels
• Integrating scattering and deep features
• Dividing clean and noisy label samples

Innovation

• Multi-model fusion integrates scattering and deep features
• GMMs divide clean and noisy label samples
• Semi-supervised learning with joint distribution alignment