EvoAI Hardware

The research is primarily focused on two areas: (1) The design, modeling, and implementation of hardware accelerators for AI methods (deep neural networks, genetic programming, etc.). (2) The use of AI, especially evolutionary algorithms, for the automated design of computer circuits such as arithmetic circuits. Inspired by nature and supported by AI, we develop highly optimized hardware and software. The members of the group are internationally recognized researchers in the fields of nature-inspired methods and models (genetic and evolutionary algorithms, neurocomputing and cellular computing), digital circuit design, FPGA technology, and approximate computing. Applications of these methods include automated logic synthesis and circuit optimization, adaptive hardware, automated neural network design, and optimization in the areas of image processing, security, transportation, and engineering.

Research interests

  • bio-inspired hardware and software
  • modeling, design and implementation of neural network accelerators
  • approximate computing
  • neuroevolution 
  • evolutionary logic synthesis and optimization
  • evolvable hardware in FPGAs
  • cellular automata and their applications
  • applications of evolutionary algorithms in computer engineering

Current research topics

Evolutionary design, optimization and approximation of digital circuits, neural architecture search, hardware accelerators for neural networks, evolvable hardware in FPGAs, evolution of cellular automata, genetic programming. 

Significant results

  • Best Student Paper Award - DDECS 2022 (J. Klhůfek, V. Mrázek)
  • Best Paper Award - DDECS 2021 (Z. Vašíček)
  • Czech head - Doctorandus of Technical Science (V. Mrázek, 2019)
  • Czech Science Foundation President Award 2017 for excellent results of the Advanced Methods for Evolutionary Design of Complex Digital Circuits project (press info, youtube video)
  • 1 Gold, 3 Silver and 1 Bronze medals from the Humies competitions annually organized at GECCO conference series.
  • Best paper/presentation awards on international conferences: DDECS 2022 (student), DDECS 2021, DATE 2017, ECTA 2016, EuroGP 2015, AHS 2013, FC 2009, MEMICS 2009, EvoHOT 2005, ICES 2003, DDECS 2002.
  • Joseph Fourier Prize for research in computer science (1st place, Z. Vašíček, 2011)
  • The Prize of Antonín Svoboda for the Best Dissertation Thesis of 2018 (V. Mrázek)
  • Czech patent no. 304181 (2013) - image filter that was automatically designed by evolutionary algorithm
  • EvoApproxLib 
  • Best paper award - DDECS 2024 (J. Klhůfek, M. Šafář, V. Mrázek, Z. Vašíček)

Cooperation

Current:

  • NYU Abu Dhabi (Prof. Muhammad Shafique, dr. Alberto Marchisio)
  • University of Patras (Prof. Georgios Zervakis)
  • University of Alberta (Prof. Jie Han)
  • University of York (Prof. Steve S. Smith)

Former:

  • Purdue University (Prof. Kaushik Roy)
  • NASA JPL Advanced Robotic Controls Group (dr. Adrian Stoica)
  • University of Oslo (Prof. Jim Torresen, Dr. Kyrre Glette)
  • Center of Industrial Electronics, Universidad Politécnica de Madrid (dr. Ruben Salvador, dr. Eduardo de la Torre)
  • Universidad Carlos III de Madrid (Antonio J. Sanchez-Clemente, dr. Luis Entrena)
  • Faculty of Informatics, Masaryk University (dr. Petr Švenda, prof. Vashek Matyáš)
  • Faculty of Information Technology, Czech Technical University in Prague (dr. Petr Fišer, dr. Jan Schmidt)

Information

Lukáš Sekanina, sekanina@fit.vutbr.cz
Web presentation: https://ehw.fit.vutbr.cz/

Back to top