Approximate computing (AC) is an emerging computing paradigm that reduces output quality for the benefit of savings in resources usage, e.g., area, energy. Recent approaches for quality assurance in AC designs use machine learning (ML) based design selectors that select the most suitable design from a given library. Nonetheless, this idea of quality assurance in AC design could be improved in many aspects. In this project, we first extend an exiting ML-based library of AC designs with more realistic models. Furthermore, we propose a novel Design Space Exploration technique that drastically reduces the AC design selection time. This technique uses mathematical modeling to assess output quality of AC designs rather than excessive simulation. In another effort in improving quality assurance for AC design, we introduce an approach for design reliability, which uses the natural characteristics of AC circuits to form a quadruple redundant modular design.
Thesis
Alain Aoun, "
On the Improving of Approximate Computing Quality Assurance", MASc Thesis, Department of Electrical and Computer Engineering, Concordia University, Montreal, Quebec, Canada, May 2021.
Conference Papers
[1] M. Masadeh, A. Aoun, O. Hasan, S. Tahar.
Highly-Reliable Approximate Quadruple Modular Redundancy with Approximation-Aware Voting. Proc. IEEE International Conference on Microelectronics (ICM'20), Irbid, Jordan, December 2020, pp. 1-4.
[2] M. Masadeh, A. Aoun, O. Hasan, S. Tahar.
Decision Tree-based Adaptive Approximate Accelerators for Enhanced Quality. Proc. IEEE International Systems Conference (SysCon'20), Montreal, Quebec, Canada, August 2020, pp. 1-5.
