Dr. Jakob Hoydis, Ph.D.

Principal Research Scientist, NVIDIA Corporation, Paris, France

Jakob Hoydis is a Principal Research Scientist at NVIDIA, specializing in the intersection of machine learning and wireless communications. Previously, he led a research department at Nokia Bell Labs in France. He holds a diploma in electrical engineering from RWTH Aachen University, Germany, and a Ph.D. from Supéléc, France. Dr. Hoydis is an IEEE Fellow and an 2023-24 Distinguished Industry Speaker for the IEEE Signal Processing Society. From 2019 to 2021, he chaired the IEEE COMSOC Emerging Technology Initiative on Machine Learning for Communications and served as an Editor for the IEEE Transactions on Wireless Communications. During this period, he was also an Area Editor for the IEEE JSAC Series on Machine Learning in Communications and Networks. Dr. Hoydis has received numerous awards, including the 2019 VTG IDE Johann-Philipp-Reis Prize, the 2019 IEEE SEE Glavieux Prize, the 2018 IEEE Marconi Prize Paper Award, and the 2015 IEEE Leonard G. Abraham Prize. Additionally, he was honored with the 2018 Nokia AI Innovation Award and the Nokia France Top Inventor Awards in 2018 and 2019. He is a co-author of the textbook “Massive MIMO Networks: Spectral, Energy, and Hardware Efficiency” (2017) and one of the maintainers and core developers of Sionna, a GPU-accelerated open-source link level simulator for next-generation communication systems.

Abstract: Modern AI advancements are driven by automatic differentiation frameworks that enable the hardware-accelerated computation of gradients of intricate functions such as large neural networks. We will explore the application of such frameworks in building simulators of communication systems that are differentiable with respect to the parameters of their components, ranging from signal processing blocks to models for radio wave propagation. This capability enables not only the end-to-end optimization of communication systems for specific environments but also the calibration of these simulators using measured data. The latter aspect is essential for the development of digital twin networks, which will form the foundation of future communication technology.