Université de Nantes

Recent developments in deep learning (DL) are putting a lot of pressure and pushing the demand for intelligent edge devices capable of on-site learning. The realization of such systems is, however, a massive challenge due to the limited resources available in an embedded context and the massive training costs for state-of-the-art deep neural networks (DNNs). In order to realize the full potential of deep learning, it is imperative to improve existing network training methodologies and the hardware being used.

In the context of the upcoming LeanAI project, funded through the Labex CominLabs initiative, we are looking for two excellent PhD candidates to work on these problems. We will be focusing on both the arithmetic and algorithmic levels with the end goal of designing new mixed numerical precision hardware architectures for DNN training that are at the same time more energy-efficient and capable of improved performance in a resource-restricted environment. The expected outcomes include new mixed-precision algorithms for neural network training, together with open-source tools for hardware and software training acceleration at the arithmetic level on edge devices. Such tools are important to help democratize access to AI technologies.

One PhD position is located at the University of Nantes and LS2N, and the other at the University of Rennes and IRISA.

Find the full job offer description and application requirements here. Applications are received until the positions are filled.

For more information contact Anastasia Volkova (anastasia.volkova@univ-nantes.fr) and Silviu-Ioan Filip (silviu.filip@inria.fr).

Pour postuler, envoyez votre CV et votre lettre de motivation par e-mail à Anastasia.volkova@univ-nantes.fr