Mark Deutel

Mark Deutel, M. Sc.

Department of Computer Science
Chair of Computer Science 12 (Hardware-Software-Co-Design)

Room 02.128
Cauerstr. 11
91058 Erlangen

Since 05/2022	Researcher at the Department of Computer Science 12 (Hardware/Software Co-Design)
2019 – 2022	Computer Science M.Sc. at Friedrich-Alexander-Universität Erlangen-Nürnberg
2015 – 2019	Computer Science B.Sc. at Friedrich-Alexander-Universität Erlangen-Nürnberg

Multi-Objective Optimization of Deep Neural Networks (DNNs) using Bayesian Optimization
- Design space exploration of DNN hyper- and compression parameters for optimized deployment on microcontrollers
- Neural Architecture Search (NAS), scalable and efficient architectures
On-Device Continual Learning for Microcontrollers
- (Class-)Incremental Learning
- Lifelong Learning
TinyML – Machine Learning on the Edge
- Efficient DNN Inference on (Cortex-M) microcontroller targets
- DNN Pruning & Quantization
- Demos and use-cases demonstrating the potential of embedded AI

Demo of MicroYOLO person detection on a STM32H743VI ARM Cortex-M7 based microcontroller (OpenMV H7 R2)

Demo of On-Device DNN Training (Animal Classification) on a STM32H743VI ARM Cortex-M7 based microcontroller (OpenMV H7 R2)

SEMESTER

Deutel M., Hannig F., Mutschler C., Teich J.:
Fused-Layer CNNs for Memory-Efficient Inference on Microcontrollers
Workshop on Machine Learning and Compression @ NeurIPS 2024 (Vancouver Convention Center, 10. December 2024 - 15. December 2024)
BibTeX: Download
Deutel M., Hannig F., Mutschler C., Teich J.:
On-Device Training of Fully Quantized Deep Neural Networks on Cortex-M Microcontrollers
In: IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (2024)
ISSN: 0278-0070
DOI: 10.1109/TCAD.2024.3484354
URL: https://ieeexplore.ieee.org/document/10726519
BibTeX: Download
Witt N., Deutel M., Sobel C., Schubert J., Woller P.:
Energy-Efficient AI on the Edge
In: Christopher Mutschler, Christian Münzenmayer, Norman Uhlmann, Alexander Martin (ed.): Unlocking Artificial Intelligence: From Theory to Applications, Springer Link, 2024, p. 359 - 380
DOI: 10.1007/978-3-031-64832-8_19
BibTeX: Download

Deutel M., Mutschler C., Teich J.:
microYOLO: Towards Single-Shot Object Detection on Microcontrollers
4th Workshop on IoT, Edge, and Mobile for Embedded Machine Learning (ITEM) (Torino, 18. September 2023 - 22. September 2023)
Open Access: https://arxiv.org/abs/2408.15865
BibTeX: Download
Deutel M., Woller P., Mutschler C., Teich J.:
Energy-efficient Deployment of Deep Learning Applications on Cortex-M based Microcontrollers using Deep Compression
Methoden und Beschreibungssprachen zur Modellierung und Verifikation von Schaltungen und Systemen MBMV’23 (Freiburg, 23. March 2023 - 24. March 2023)
In: VDE (ed.): MBMV 2023; 26th Workshop 2023
Open Access: https://arxiv.org/abs/2205.10369
URL: https://ieeexplore.ieee.org/document/10173060
BibTeX: Download