Publication Details

A Fast Design Space Exploration Framework for the Deep Learning Accelerators: Work-in-Progress

COLUCCI Alessio, MARCHISIO Alberto, BUSSOLINO Beatrice, MRÁZEK Vojtěch, MARTINA Mauricio, MASERA Guido and SHAFIQUE Muhammad. A Fast Design Space Exploration Framework for the Deep Learning Accelerators: Work-in-Progress. In: 2020 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)}. Singapore: Institute of Electrical and Electronics Engineers, 2020, pp. 34-36. ISBN 978-1-7281-9198-0.
Czech title
Rychlý framework pro prohledávání a návrh konfigurací akcelerátorů hlubokých neuronových sítí
Type
conference paper
Language
english
Authors
Colucci Alessio (TU-Wien)
Marchisio Alberto (TU-Wien)
Bussolino Beatrice (POLITO)
Mrázek Vojtěch, Ing., Ph.D. (DCSY FIT BUT)
Martina Mauricio (POLITO)
Masera Guido (POLITO)
Shafique Muhammad (TU-Wien)
Keywords

Design Space Exploration, Hardware Accelerator, Capsule Networks, Convolutional Neural Networks, Training

Abstract

The Capsule Networks (CapsNets) is an advanced form of Convolutional Neural Network (CNN), capable of learning spatial relations and being invariant to transformations. CapsNets requires complex matrix operations which current accelerators are not optimized for, concerning both training and inference passes. Current state-of-the-art simulators and design space exploration (DSE) tools for DNN hardware neglect the modeling of training operations, while requiring long exploration times that slow down the complete design flow. These impediments restrict the real-world applications of CapsNets (e.g., autonomous driving and robotics) as well as the further development of DNNs in life-long learning scenarios that require training on low-power embedded devices. Towards this, we present XploreDL , a novel framework to perform fast yet high-fidelity DSE for both inference and training accelerators, supporting both CNNs and CapsNets operations. XploreDL enables a resource-efficient DSE for accelerators, focusing on power, area, and latency, highlighting Pareto-optimal solutions which can be a green-lit to expedite the design flow. XploreDL can reach the same fidelity as ARM's SCALE-sim, while providing 600x speedup and having a 50x lower memory-footprint. Preliminary results with a deep CapsNet model on MNIST for training accelerators show promising Pareto-optimal architectures with up to 0.4 TOPS/squared-mm and 800 fJ/op efficiency. With inference accelerators for AlexNet the Pareto-optimal solutions reach up to 1.8 TOPS/squared-mm and 200 fJ/op efficiency.

Published
2020
Pages
34-36
Proceedings
2020 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)}
Conference
2020 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS), Singapore, SG
ISBN
978-1-7281-9198-0
Publisher
Institute of Electrical and Electronics Engineers
Place
Singapore, SG
DOI
UT WoS
000833313100014
EID Scopus
BibTeX
@INPROCEEDINGS{FITPUB12420,
   author = "Alessio Colucci and Alberto Marchisio and Beatrice Bussolino and Vojt\v{e}ch Mr\'{a}zek and Mauricio Martina and Guido Masera and Muhammad Shafique",
   title = "A Fast Design Space Exploration Framework for the Deep Learning Accelerators: Work-in-Progress",
   pages = "34--36",
   booktitle = "2020 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)\}",
   year = 2020,
   location = "Singapore, SG",
   publisher = "Institute of Electrical and Electronics Engineers",
   ISBN = "978-1-7281-9198-0",
   doi = "10.1109/CODESISSS51650.2020.9244038",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/12420"
}
Back to top