QoE Optimization for Wireless Multimedia Communications
摘要
Recent works have primarily focused on semantic communications for text/speech modalities, thus falling under the category of text semantic communications (TSC). In (Farsad et al. 2018), a deep learning–based joint source-channel coding (JSCC) approach for text is proposed, which achieves a lower word error rate and preserves semantic information of sentences. Later, in Lu et al. (2021), the authors design a reinforcement learning (RL)–based semantic communication system to handle the non-differentiability of semantic metrics and interact with the surrounding noisy environment.