3D integration of scalable and autonomous Photonic Neural Networks
Lecturer: Daniel Brunner - Institut FEMTO-ST, Université Bourgogne Franche-Comté CNRS UMR 6174, Besançon, France
Date: Jan. 19, 2024 11:30 a.m. - 1 p.m.
Location: Querzoli
Organizer: Wiersma Diederik

Neural network (NN) concepts revolutionize computing by solving challenges previously thought to be reserved to the abstract intelligence of humans. However, the astonishing and substantial conceptual breakthroughs are so far not mirrored by advances in integrated hardware specialized in physically implementing NNs. As always with computing, scalability is the key metric. Integrated photonic architectures have the potential to revolutionize energy consumption and speed. However, conventional 2D lithography strongly limits the size of integrated NNs due to fundamental scaling laws. We want to overcome this problem by using 3D printed photonic integration, where photonic waveguides realizing a NN’s connections. Finally, for maximal efficiency, the largest fraction of a NN’s hardware should be dedicated to the core computational task, while auxiliary infrastructure should be pushed into the background. I will demonstrate a fully autonomous photonic NN based on a high-dimensional semiconductor laser that implements a scalable photonic NN fully in parallel and with minimal support by a classical digital computer.