Presentation title

ALOHA: a software framework for runtime-Adaptive and secure deep Learning On Heterogeneous Architectures

Authors

Paolo Meloni, Gianfranco Deriu, Daniela Loi and Luigi Raffo

Institution(s)

Università degli Studi di Cagliari, Italy

Presentation type

Technical presentation

Abstract

This work proposes a framework for supporting the implementation of Deep Learning (DL) algorithms on heterogeneous low-energy computing platforms. The method allows automating i) the selection of an optimal algorithm configuration, ii) the optimization of its partitioning and mapping on a target processing platform, and iii) the optimization of power and energy savings during its deployment. The approach will been practically validated on NEURAghe, a flexible and efficient hardware/software solution for the acceleration of CNNs on Zynq SoCs, showing that it can actually be supported by state-of-the-art technologies.

---

Additional material

• Presentation slides: [pdf]

For more details on this presentation please click the button below: