Project Details

AppNeCo: Aproximativní neurovýpočty

Project Period: 1. 1. 2022 - 31. 12. 2024

Project Type: grant

Code: GA22-02067S

Agency: Czech Science Foundation

Program: Standardní projekty

English title
AppNeCo: Approximate Neurocomputing
Type
grant
Keywords

approximate computing,convolutional networks,energy complexity,robust learning,hardware accelerator,image classification

Abstract

Nowadays, modern AI technologies based on deep neural networks, whose computation is demanding on energy consumption, are implemented in devices with limited resources (e.g. battery powered cellphones). In error-tolerant applications (e.g. image classification), the use of approximate computing methods can save enormous amount of energy at the cost of only a small loss in accuracy. AppNeCo is a basic research project of approximate neurocomputing, whose ambition is an original synergy of approximation and complexity theory of neural networks and empirical experience with the top design of high-performance approximate implementations of hardware circuits. Its goal is to develop complexity-theoretic foundations of approximate computation by convolutional neural networks (CNN) of bounded energy complexity for application domains specified by input space distributions. This knowledge will be used in designing new strategies for approximating components and learning algorithms of low-energy high-precision CNNs. The new methods will be tested on image processing tasks.

Team members
Sekanina Lukáš, prof. Ing., Ph.D. (UPSY FIT VUT) , research leader
Klhůfek Jan, Ing. (UPSY FIT VUT)
Mrázek Vojtěch, Ing., Ph.D. (UPSY FIT VUT)
Vašíček Zdeněk, doc. Ing., Ph.D. (UPSY FIT VUT)
Publications

2024

  • VAŠÍČEK Zdeněk. Automated Synthesis of Commutative Approximate Arithmetic Operators. In: 2024 IEEE Congress on Evolutionary Computation, CEC 2024 - Proceedings. Yokohama: IEEE Computer Society, 2024, pp. 1-8. ISBN 979-8-3503-0836-5. Detail
  • ŠÍMA Jiří, MRÁZEK Vojtěch and VIDNEROVÁ Petra. Energy Complexity of Convolutional Neural Networks. Neural Computation, vol. 36, no. 8, 2024, pp. 1601-1625. ISSN 0899-7667. Detail
  • MRÁZEK Vojtěch, KOKKINIS Argyrios, PAPANIKOLAOU Panagiotis, VAŠÍČEK Zdeněk, SIOZIOS Kostas, TZIMPRAGOS Georgios, TAHOORI Mehdi and ZERVAKIS Georgios. Evolutionary Approximation of Ternary Neurons for On-sensor Printed Neural Networks. In: 2024 IEEE/ACM International Conference on Computer Aided Design (ICCAD). New Jersey, 2024, p. 9. Detail

2023

  • ŠÍMA Jiří, VIDNEROVÁ Petra and MRÁZEK Vojtěch. Energy Complexity Model for Convolutional Neural Networks. In: Artificial Neural Networks and Machine Learning - ICANN 2023: 32nd International Conference on Artificial Neural Networks. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Heraklion, 2023, pp. 186-198. ISBN 978-3-031-44203-2. Detail
  • SEDLÁČEK Marek and SEKANINA Lukáš. Evolution of Editing Scripts From Examples. In: Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO '23). Lisbon: Association for Computing Machinery, 2023, pp. 803-806. ISBN 979-8-4007-0120-7. Detail
  • KALKREUTH Roman, VAŠÍČEK Zdeněk, HUSA Jakub, VERMETTEN Diederick, YE Furong and THOMAS Bäck. General Boolean Function Benchmark Suite. In: FOGA 2023 - Proceedings of the 17th ACM/SIGEVO Conference on Foundations of Genetic Algorithms. Potsdam: Association for Computing Machinery, 2023, pp. 84-95. ISBN 979-8-4007-0202-0. Detail
  • PIŇOS Michal, MRÁZEK Vojtěch and SEKANINA Lukáš. Prediction of Inference Energy on CNN Accelerators Supporting Approximate Circuits. In: 2023 26th International Symposium on Design and Diagnostics of Electronic Circuits and Systems. Talinn: Institute of Electrical and Electronics Engineers, 2023, pp. 45-50. ISBN 979-8-3503-3277-3. Detail
  • KALKREUTH Roman, VAŠÍČEK Zdeněk, HUSA Jakub, VERMETTEN Diederick, YE Furong and THOMAS Bäck. Towards a General Boolean Function Benchmark Suite. In: GECCO 2023 Companion - Proceedings of the 2023 Genetic and Evolutionary Computation Conference Companion. New York: Association for Computing Machinery, 2023, pp. 591-594. ISBN 979-8-4007-0120-7. Detail

2022

  • KLHŮFEK Jan and MRÁZEK Vojtěch. ArithsGen: Arithmetic Circuit Generator for Hardware Accelerators. In: 2022 25th International Symposium on Design and Diagnostics of Electronic Circuits and Systems (DDECS '22). Prague: Institute of Electrical and Electronics Engineers, 2022, pp. 44-47. ISBN 978-1-6654-9431-1. Detail
  • KOCNOVÁ Jitka and VAŠÍČEK Zdeněk. Delay-aware evolutionary optimization of digital circuits. In: Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI. Nicosia, Cyprus: IEEE Computer Society, 2022, pp. 188-193. ISBN 978-1-6654-6605-9. Detail
  • VÁLEK Matěj and SEKANINA Lukáš. Evolutionary Approximation in Non-Local Means Image Filters. In: 2022 IEEE International Conference on Systems, Man, and Cybernetics (SMC). Praha: Institute of Electrical and Electronics Engineers, 2022, pp. 2759-2766. ISBN 978-1-6654-5258-8. Detail
  • MARCHISIO Alberto, MRÁZEK Vojtěch, MASSA Andrea, BUSSOLINO Beatrice, MARTINA Mauricio and SHAFIQUE Muhammad. RoHNAS: A Neural Architecture Search Framework with Conjoint Optimization for Adversarial Robustness and Hardware Efficiency of Convolutional and Capsule Networks. IEEE Access, vol. 2022, no. 10, pp. 109043-109055. ISSN 2169-3536. Detail
Back to top