移动通信2025，Vol.49Issue(7)：55-60,6.DOI:10.3969/j.issn.1006-1010.20250603-0001
高斯源下率失真分类函数的通用表达
Universal Representation of Rate-Distortion Classification Function for Gaussian Sources
摘要
Abstract
This paper investigates the rate-distortion classification function and its universal representation for Gaussian sources.To address the limitations of traditional rate-distortion theory,a rate-distortion-perception-classification model integrating classification constraints and perception requirements is proposed.The closed-form solution for the rate-distortion classification function under scalar Gaussian sources is derived,demonstrating that the maximum distortion corresponding to the optimal classification constraint is twice the minimum distortion.Additionally,a universal framework is introduced,where a single encoder satisfies multiple constraints without a penalty for rate distortion.The results indicate a fundamental trade-off between rate,distortion,and classification performance,providing theoretical support for semantic compression in machine communication.关键词
率失真分类函数/高斯信源/码率-失真-分类权衡/通用表示/信息理论Key words
rate distortion classification function/gaussian source/rate-distortion-classification trade-off/universal representation/information theory分类
信息技术与安全科学引用本文复制引用
唐浩文,施雨轩,吴泳澎..高斯源下率失真分类函数的通用表达[J].移动通信,2025,49(7):55-60,6.
