| KEYWORD |
Deep neural network synthesis for FPGAs via high-level synthesis
Thesis in external company 
Thesis abroad
keywords HIGH-LEVEL SYNTHESIS, MACHINE LEARNING
Reference persons LUCIANO LAVAGNO, MIHAI TEODOR LAZARESCU
External reference persons Fabian Chersi, CEA Paris Saclay
Research Groups Microelectronics
Description The goal of the thesis is to prototype and test the code generation capabilities of the AIdge ML deployment framework for FPGAs.
The tasks will be:
- select a suitable CNN for the experiment (perhaps Resnet20)
- define a list of layers that need to be supported (probably 2-3)
- write the c++ code implementing the layers (probably about 1000 lines of code)
- use the generated code in Vitis HLS and analyze the results
Required skills C/C++
Digital HW design
Knowledge of HLS is preferred
Notes The thesis can be carried out at CEA Saclay or at Politecnico di Torino, under the guidance of both groups, based on the student's preferences
Deadline 31/10/2025
PROPONI LA TUA CANDIDATURA