Skip to main content


Hardware and Software architectures for deep learning acceleration on embedded multi-processor

Hardware processing, Computer vision, Artificial intelligence

The general topic of this thesis is hardware and software infrastructure for the efficient execution of neural networks. In particular, the objective is to study the interesting strategies to equip a family of processors with hardware accelerators associated with a software architecture for the efficient execution of neural networks.
The main issues are related to the ability to ship the smart node or nodes in SoCs that are highly constrained in terms of area, energy consumption and available computing power.


Thesis director: Stéphane MANCINI
Thesis supervisor: Stefanie HAHMANN (LJK)
Thesis started on: March 2018
Thesis defence: 04/01/2022
Doctoral school: MSTII

Submitted on January 12, 2022

Updated on February 9, 2022