< Back to previous page
Researcher
Koen Goetschalckx
- Disciplines:Artificial intelligence not elsewhere classified, Digital integrated circuits, Pattern recognition and neural networks
Affiliations
- Electronic Circuits and Systems (ECS) (Division)
Member
From1 Aug 2020 → Today - ESAT - MICAS, Microelectronics and Sensors (Division)
Member
From1 Sep 2017 → 31 Jul 2020
Projects
1 - 1 of 1
- Optimizing Embedded Vision Efficiency across the Hardware-dataflow-algorithm StackFrom1 Sep 2017 → 26 Apr 2023Funding: Own budget, for example: patrimony, inscription fees, gifts
Publications
1 - 7 of 7
- DepFiN: A 12-nm Depth-First, High-Resolution CNN Processor for IO-Efficient Inference(2023)
Authors: Koen Goetschalckx, Marian Verhelst
- Optimizing Embedded Vision Efficiency across the Hardware-dataflow-algorithm Stack(2023)
Authors: Koen Goetschalckx, Marian Verhelst
- DepFiN: A 12nm, 3.8TOPs depth-first CNN processor for high res. image processing(2021)
Authors: Koen Goetschalckx, Marian Verhelst
Pages: 1 - 2 - Breaking High-Resolution CNN Bandwidth Barriers With Enhanced Depth-First Execution(2019)
Authors: Koen Goetschalckx, Marian Verhelst
Pages: 323 - 331 - Optimized Hierarchical Cascaded Processing(2018)
Authors: Koen Goetschalckx, Steven Lauwereins, Martin Andraud, Marian Verhelst
Pages: 884 - 894 - Efficiently Combining SVD, Pruning, Clustering and Retraining for Enhanced Neural Network Compression(2018)
Authors: Koen Goetschalckx, Patrick Wambacq, Marian Verhelst
Number of pages: 6 - Minimum Energy Quantized Neural Networks(2017)
Authors: Bert Moons, Koen Goetschalckx, Marian Verhelst
Pages: 1921 - 1925Number of pages: 5