Topic: Semantic data fusion in wireless networks of intelligent machines
Position: PhD / CIFRE
Institution: Advanced Wireless Technology Labs, Huawei Paris Research Centre
Context: Communication between intelligent machines is a major aspect of future wireless networks. While current networks are designed to support the needs of humans, communication between machines exhibits largely different protocols and requirements. The design of mechanisms and networking for such communication lies in the intersection of machine learning, wireless networks, and semantic communications.
Deep neural networks show internal logical behaviour and capabilities to extract semantic meaning, as shown in ongoing mathematical analysis using category-theory analysis in our team. Such semantic information and semantic communication is expected to play a key role in efficient machine-to-machine communications.
Wireless channels are characterised by time variance and multiple access. Time-variance makes connectivity and reliability partly uncontrollable, and multiple-access requires user scheduling or other multiple-access strategies. These aspects pose particular challenges for communication between machines.
PhD project: Data fusion is a specific scenario and application within the general picture of communication between intelligent machines. Devices observe and process local data, and communicate this over a joint wireless channel to a centre that fuses this data. The may be seen as a distributed machine learning for classification or regression. The time-varying and resource-limited wireless channel imposes specific constraints and challenges on the design and operation of such a system. On the other hand, goal orientation and semantic aspects allow for and demand specific optimisation.
The PhD project aims at the mathematical analysis and practical implementation of such a data fusion system. The mathematical aspects may include fusion based on sheaf theory, exploitation of logical and semantic analysis of neural networks, and utilisation of the topological structure of the data.
Requirements: The applicant is expected to have the following skills:
Application: The position is open now. Please send your full CV, motivation letter, and contact information for two references or letters of reference.
Contact: For questions and your application, contact Dr Ingmar Land, Huawei Paris Research Centre, by email to firstname.lastname@example.org.
(c) GdR 720 ISIS - CNRS - 2011-2020.