重庆邮电大学学报（自然科学版）2026，Vol.38Issue(1)：65-73,9.DOI:10.3979/j.issn.1673-825X.202501150018
基于监督判别学习的图像抗噪语义通信机制
Image anti-noise semantic communication mechanism based on supervised discriminant learning
摘要
Abstract
Semantic communication refers to the process of understanding,interpreting,and processing information by exploiting semantic meaning during communication.Compared with traditional communication paradigms,it not only improves transmission efficiency and reliability but also significantly reduces the time required for information exchange and comprehension.However,semantic communication systems are highly vulnerable to semantic noise,which can severely degrade system performance.Effectively mitigating noise interference during semantic communication remains a challenging problem.To address this issue,an image semantic communication mechanism based on supervised discriminative learning is proposed.Simulation results on benchmark datasets demonstrate that,compared with baseline algorithms,the proposed method achieves notable performance improvements under both semantic noise and physical channel noise,while simultane-ously ensuring accurate data reconstruction and reliable task execution.关键词
语义噪声/语义通信/对抗训练/深度监督网络Key words
semantic noise/semantic communication/adversarial training/deep supervision network分类
信息技术与安全科学引用本文复制引用
应俊,徐伟伟,祝志远,周继华,王汝言,吴大鹏..基于监督判别学习的图像抗噪语义通信机制[J].重庆邮电大学学报（自然科学版）,2026,38(1):65-73,9.基金项目
国家自然科学基金项目(U24A20211,62271096,U20A20157) （U24A20211,62271096,U20A20157）
重庆市自然科学基金创新发展联合基金项目(CSTB2023NSCQ-LZX0134,CSTB2024NSCQ-LZX0124) （CSTB2023NSCQ-LZX0134,CSTB2024NSCQ-LZX0124）
重庆市高校创新研究群体项目(CXQT20017) （CXQT20017）
重邮信通青创团队支持计划项目(SCIE-QN-2022-04) National Natural Science Foundation of China(U24A20211,62271096,U20A20157) （SCIE-QN-2022-04）
Natural Science Founda-tion of Chongqing,China(CSTB2023NSCQ-LZX0134,CSTB2024NSCQ-LZX0124) （CSTB2023NSCQ-LZX0134,CSTB2024NSCQ-LZX0124）
University Innovation Research Group of Chongqing(CXQT20017) （CXQT20017）
Youth Innovation Group Support Program of ICE Discipline of CQUPT(SCIE-QN-2022-04) （SCIE-QN-2022-04）
