Ahmet Caner Yüzügüler

Ph.D. Candidate at EPFL

Lausanne, Switzerland








Deep learning Hardware accelerators FPGA/ASIC design Computer architecture Embedded systems

Programming Languages

C/C++ Python Pytorch/Tensorflow VHDL/Verilog CUDA

Design Tools

Xilinx ISE/Vivado Design Tools Altera Quartus Modelsim Matlab Simulink


École Polytechnique Fédérale de Lausanne (EPFL)
Lausanne, Switzerland
2018 - Current

Ph.D. in Computer Science

Advisor: Prof. Pascal Frossard

École Polytechnique Fédérale de Lausanne (EPFL)
Lausanne, Switzerland
2015 - 2018

M.Sc. in Computer Science

Advisor: Prof. Pascal Frossard
Middle East Technical University
Ankara, Turkey
2009 - 2014

B.Sc. in Electrical Engineering

Work Experience

Doctoral Assistant - EPFL / Parallel Systems Architecture Laboratory (PARSA)
Lausanne, Switzerland
Sept. 2018 - Current

  • Developed a neural architecture search framework that optimizes deep neural networks for target hardware accelerators.
  • Developed a novel scale-out systolic array architecture that increases hardware utilization and power efficiency in DNN accelerators.
  • Developed a novel analog circuit that processes deep neural networks more efficiently. Published [1].
  • Implemented convolutional neural networks on the integrated FPGA of an Intel HARP platform.

Research Intern - ABB Corporate Research
Baden, Switzerland
Feb. 2017 - March 2018

  • Worked on my master's thesis on the subject of approximating system models with deep neural networks for real-time simulations. Published [2].

Research Assistant - EPFL / Rigorous System Design Laboratory
Lausanne, Switzerland
Sept. 2015 - August 2016

  • Worked on the algorithm design and software implementation of a 3D ultrasound application on an FPGA and a many-core Kalray MPPA-256 platform. Published [3], [4].

Hardware Design Engineer - Aselsan Inc. / Division of Defense Systems Technologies
Ankara, Turkey
July 2014 - August 2015

  • Worked on the digital hardware design of embedded computers for defense applications. Published [5].


1     U-Boost NAS: Utilization-Boosted Differentiable Neural Architecture Search,
A. C. Yüzügüler, N. Dimitriadis, P. Frossard
European Conference on Computer Vision (ECCV), 2022

2     Scale-out Systolic Arrays
A. C. Yüzügüler, C. Sönmez, M. Drumond, Y. Oh, B. Falsafi, P. Frossard
Preprint, 2022

3     Equinox: Training (for Free) on a Custom Inference Accelerator
M. Drumond, L. Coulon, A. Pourhabibi, A. C. Yüzügüler, B. Falsafi, M. Jaggi
MICRO-54: 54th Annual IEEE/ACM International Symposium on Microarchitecture, 2021

4     Analog Neural Networks with Deep-submicrometer Nonlinear Synapses
A. C. Yüzügüler, F. Celik, M. Drumond, B. Falsafi, P. Frossard
IEEE Micro Special Issue on Machine Learning Acceleration, 2019

5     Towards Commoditizing Simulations of System Models Using Recurrent Neural Networks
A. C. Yüzügüler, A. Moga, C. Franke
IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm), 2018

6     1024-Channel 3D ultrasound digital beamformer in a single 5W FPGA
F. Angiolini, A. Ibrahim, W. Simon, A. C. Yüzügüler, M. Arditi, J. P. Thiran and G. De Micheli
Design, Automation & Test in Europe Conference & Exhibition (DATE), 2017

7     (Demo) Single-FPGA 3D Ultrasound Beamformer
A. C. Yüzügüler, W. Simon, A. Ibrahim, F. Angiolini, M. Arditi, J. P. Thiran and G. De Micheli
International Conference on Field-Programmable Logic and Applications (FPL), 2016

8     Changing Utilization Rates in Real Time to Investigate FPGA Power Behavior
A. C. Yüzügüler and E. Sahin
Xilinx Xcell Journal Issue 89, 2015

9     Transformation-invariant Dictionary Learning for Fast Image Classification
A. C. Yüzügüler, E. Vural and P. Frossard
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2014