Semantic encoders, decoders, and semantic communication frameworks

Abstract

The invention discloses a semantic encoder, a decoder and a semantic communication framework, and belongs to the field of semantic communication. Based on the traditional semantic communication architecture, the present invention adds a semantic comparator to the semantic encoder and a semantic reasoner to the semantic decoder. The semantic communication architecture does not directly encode the semantic information on the sender, but trains a semantic comparator that can distinguish the expert reasoning path of the sender. The semantic reasoner learns through adversarial training with the semantic comparator. The reasoning mechanism of the semantic sender's expert reasoning path can finally directly decode the received information and recover the semantic information, which solves the problem of difficulty in encoding the semantic reasoning mechanism in implicit semantic information and can realize the decoding of semantic information. .

Description

technical field [0001] The invention belongs to the field of semantic communication, and more specifically relates to a semantic encoder, a decoder and a semantic communication framework. Background technique [0002] Traditional communication systems are now seen as one of the biggest obstacles in delivering quality-of-experience-based services to end users. This has inspired a new communication paradigm—semantic communication. Compared with traditional communication systems, semantic communication can allow all communicators, including information sources and transmission receivers, to utilize shared syntax, semantics, and inference rules to help transmit and interpret target meanings. Recent studies have shown that semantic communication, a communication paradigm, has the potential to improve communication efficiency, reliability, and enhance user quality of experience. [0003] Existing work on semantic communication focuses on transferring explicit semantic informatio...

AI Technical Summary

Problems solved by technology

However, there are many difficulties in extracting implicit semantic information

First, there is a lack of a simple and expressive way to express implicit semantics

Images