Description du sujet de thèse

Neuro-inspired nanoelectronic systems, which mimic the functioning of the human brain, use components like memristors to simulate synapses. These systems present a promising avenue for low-energy consumption electronic systems. Memristors enable efficient implementation of artificial synapses due to their ability to integrate computation directly within storage elements, following the principle of in-memory computing. However, the use of memristors is primarily limited to inference, while their potential for learning remains largely unexplored due to numerous technical challenges. Spin-orbit torque (SOT) devices exhibit unique physical properties, particularly an intrinsic stochastic behavior, making them promising candidates for the hardware implementation of artificial synapses that can learn. While these devices have been primarily studied for non-volatile memory applications, they hold untapped potential for neuro-inspired approaches, particularly in probabilistic learning systems.

Objectives of the Project:
• Utilize SOT devices as stochastic artificial synapses: Model and simulate their physical behavior to harness their probabilistic nature.
• Study the integration of Bayesian algorithms: Leverage the stochastic characteristics of SOT devices for Bayesian learning models adapted to neural networks.
• Optimize nanodevices for learning: Develop strategies to optimize the physical properties of SOT devices to meet the specific needs of neuro-inspired applications, distinguishing them from their conventional use as memory devices.

Contexte de travail

The project will take place within the INTEGNANO team at the Centre for Nanoscience and Nanotechnology, which focuses on an integrative and multidisciplinary approach to developing new charge- and spin-based devices. We combine expertise in nanomagnetism and electronics, materials science, circuit design, and innovative computing architectures. Our research interests range from the study of fundamental phenomena to the design of novel devices with potential applications in information processing and storage.

The project is a collaboration with the University of Texas at Dallas (USA).

Le poste se situe dans un secteur relevant de la protection du potentiel scientifique et technique (PPST), et nécessite donc, conformément à la réglementation, que votre arrivée soit autorisée par l'autorité compétente du MESR.