Write a Blog >>
LCTES 2018
co-located with PLDI 2018

The recent ground-breaking advances in deep learning networks (DNNs) make them attractive for embedded systems. However, it can take a long time for DNNs to make an inference on resource-limited embedded devices. Offloading the computation into the cloud is often infeasible due to privacy concerns, high latency, or the lack of connectivity. As such, there is a critical need to find a way to effectively execute the DNN models locally on the devices. This paper presents an adaptive scheme to determine which DNN model to use for a given input, by considering the desired accuracy and inference time. Our approach employs machine learning to develop a predictive model to quickly select a pre-trained DNN to use for a given input and the optimization constraint. We achieve this by first training off-line a predictive model, and then use the learnt model to select a DNN model to use for new, unseen inputs. We apply our approach to the image classification task and evaluate it on a Jetson TX2 embedded deep learning platform using the ImageNet ILSVRC 2012 validation dataset. We consider a range of influential DNN models. Experimental results show that our approach achieves a 7.52% improvement in inference accuracy, and a 1.8x reduction in inference time over the most-capable single DNN model.

Tue 19 Jun
Times are displayed in time zone: (GMT-04:00) Eastern Time (US & Canada) change

14:00 - 15:40: LCTES 2018 - Full paper session on Adaptation and Hardware at Discovery AB
LCTES-2018-papers14:00 - 14:25
Ben TaylorLancaster University, UK, Vicent Sanz MarcoLancaster University, Willy WolffLancaster University, Yehia ElkhatibLancaster University, Zheng WangLancaster University
LCTES-2018-papers14:25 - 14:50
Lei Han, Zhaoyan ShenThe Hong Kong Polytechnic University, Zili ShaoThe Hong Kong Polytechnic University, Tao LiUniversity of Florida
LCTES-2018-papers14:50 - 15:15
Bernhard EggerSeoul National University, Eunjin SongSeoul National University, Hochan LeeSeoul National University, Daeyoung ShinSeoul National University
LCTES-2018-papers15:15 - 15:40
Michael StokesFlorida State University, Ryan BairdFlorida State University, Zhaoxiang JinMichigan Technological University, David B. Whalley, Soner OnderMichigan Technological University