面向智能机器通信的语义信息刻画及度量

Description and Measurement of Semantic Information for the Intelligent Machine Communication

语义通信中关注信息的内在含义,而非关注编码符号,有助于解决智能机器通信中超大规模连接和海量数据传输的难题,以实现机器间的实时智能通信。针对智能机器的语义通信系统,基于信息论和人工智能技术,描述了语义信息高度抽象、智能简约的特性,归纳了端到端语义通信系统的设计方法、语义信息的刻画及度量方法、不同模态语义信息的编解码方法,分析了语义通信系统的评价指标,展望了应用前景。对现有研究存在的不足提出了进一步的研究方向。

Semantic communication focuses on the meaning of information rather than encoded symbols, which is helpful to solve the problem of massive connection and data transmission in intelligent machine communication and realize real-time intelligent communication between machines. The semantic communication system with intelligent machines is stuied. First, based on information theory and artificial intelligence technology, the highly abstraction, intelligent and simple characteristics of semantic information are described, the design methods of an end-to-end semantic communication system are summarized, and the characterization and measurement methods of semantic information are presented. Then, the encoding and decoding methods of semantic information in different modes are summarized, and the evaluation indexes of a semantic communication system are analyzed. Furthermore, the research direction is given based on the shortcomings of the existing research.