Quantum-Inspired Robust and Scalable SAR Object Classification

2026-04-28

AI Summary

This work addresses the challenges of robustness and model lightweighting in synthetic aperture radar (SAR) image classification, which arise from strong noise, high dynamic range, and edge deployment constraints. For the first time, the study introduces tensor networks inspired by quantum computing to construct a lightweight classification model. By integrating robust training strategies, the approach effectively enhances model stability under data poisoning attacks and in high-noise environments. Experimental results demonstrate that the proposed method significantly reduces model size while maintaining high classification accuracy and exhibits superior resilience to interference compared to conventional neural networks.

Abstract

SAR image classification naturally has to deal with huge noise and a high dynamic range particularly requiring robust classification models. Additionally, the deployment of these models on edge devices, such as drones and military aircraft, requires a careful balance between model size and classification accuracy. This study explores the potential of tensor networks to meet these robustness requirements, specifically evaluating their resilience to data poisoning. Unlike previous works that concentrated on conventional neural networks for SAR object detection, this research focuses on the robustness and model reduction capabilities of tensor networks in object classification. Our findings indicate that tensor networks are adept at addressing both the challenges of robustness and the need for model efficiency, thereby contributing valuable insights to the ongoing discourse in radar applications and deep learning methodologies in general.

Problem

• SAR object classification
• robustness
• model efficiency
• edge deployment
• data poisoning

Innovation

• tensor networks
• SAR object classification
• robustness
• model compression
• data poisoning resilience