Electrical Engineering and Systems Science > Signal Processing
[Submitted on 30 Oct 2022]
Title:Semantic-Native Communication: A Simplicial Complex Perspective
View PDFAbstract:Semantic communication enables intelligent agents to extract meaning (or semantics) of information via interaction, to carry out collaborative tasks. In this paper, we study semantic communication from a topological space perspective, in which higher-order data semantics live in a simplicial complex. Specifically, a transmitter first maps its data into a $k$-order simplicial complex and then learns its high-order correlations. The simplicial structure and corresponding features are encoded into semantic embeddings in latent space for transmission. Subsequently, the receiver decodes the structure and infers the missing or distorted data. The transmitter and receiver collaboratively train a simplicial convolutional autoencoder to accomplish the semantic communication task. Experiments are carried out on a real dataset of Semantic Scholar Open Research Corpus, where one part of the semantic embedding is missing or distorted during communication. Numerical results show that the simplicial convolutional autoencoder enabled semantic communication effectively rebuilds the simplicial features and infer the missing data with $95\%$ accuracy, while achieving stable performance under channel noise. In contrast, the conventional autoencoder enabled communication fails to infer any missing data. Moreover, our approach is shown to effectively infer the distorted data without prior simplicial structure knowledge at the receiver, by learning extracted semantic information during communications. Leveraging the topological nature of information, the proposed method is also shown to be more reliable and efficient compared to several baselines, notably at low signal-to-noise (SNR) levels.
Current browse context:
eess.SP
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.