Our group develops NN based applications targeting resource efficient hardware platforms. To unify the mapping process from various NN frameworks to target platforms, a common intermediate representation (IR) is required.
ONNX is the new standard for exchanging neural network (NN) graphs, which becomes supported by more and more tools.
One goal of this project is the conversion of a framework specific description to ONNX. Another goal is the mapping of ONNX graphs onto our hardware platforms including Multicore and FPGA systems. This project brings together high level programming of modern NN frameworks with low level programming in C or synthesizing hardware with HLS.
In case this would not find enough interest for a team project, this project proposal would be also offered (in reduced/modified form)
[x] as individual project
[x] as project for 2-3 students
- Any of C, C++, HLS, Python3
| Module | Course | Requirements | |
|---|---|---|---|
| 39-M-Inf-GP Basic Project Intelligent Systems Grundlagenprojekt Intelligente Systeme | Gruppenprojekt | Ungraded examination
|
Student information |
The binding module descriptions contain further information, including specifications on the "types of assignments" students need to complete. In cases where a module description mentions more than one kind of assignment, the respective member of the teaching staff will decide which task(s) they assign the students.