PhD position in AI architecture optimisation

The market for Artificial Intelligence (AI) applications, particularly Artificial Neural Networks (NNs), has experienced tremendous growth in the past decade. Executing NNs on embedded multicore platforms presents a significant challenge due to resource scarcity, real-time constraints, and energy budgets. Optimizing memory and computational resource usage is crucial to meet application constraints. To enable the efficient execution of AI applications, modern platforms often include hardware accelerators and use Instruction Set Architectures (ISA) that can be substantially upgraded to offer optimizations for the processing of AI.

In this context, we aim at proposing a hybrid modeling approach to offer fast yet accurate evaluation of NN implementation on heterogeneous multi-core platforms with possible ISA extensions and HW accelerator. This flow takes as inputs NN hyperparameters, the specified multi-core platform, and user-defined constraints (e.g. maximum energy budget). Estimations are obtained through the use of system-level simulation, allowing accurate modeling of shared resource effects. Timing, power, and memory use models are characterized through measurements performed on a real platform, providing highly accurate estimations. Due to possible variations in timing, probabilistic modeling will be considered to appropriately capture and provide confident prediction of external memory and caches effects.

This proposal subscribes in an ongoing collaboration between the IETR lab in Nantes, France and the German Aerospace Space (DLR for Deutsches Zentrum f¨ur Luft - und Raumfahrt e.V.) institute Systems Engineering for future mobility (SE) in Oldenburg, Germany, previously OFFIS e.V.

This work will be conducted as a three-year PhD project, with the PhD student spending 18 months at IETR in Nantes, France, and 18 months at DLR SE in Oldenburg.