Delivering Deep Learning to Mobile Devices via Offloading

Ran, Xukan; Chen, Haoliang; Liu, Zhenming; Chen, Jiasi

doi:10.1145/3097895.3097903

Cited by 73 publications

(25 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Convolutional neural networks (CNNs) is a member of the artificial neural network (ANN) family, which is widely applied on mobile AR applications [14]. Researchers utilize deep learning on user behavior analytics [15] and mobile AR system for object recognition and context-aware tracking [16].…”

Section: B Deep Learning and Mobile Ar Using Semg And Imumentioning

confidence: 99%

MyoKey: Surface Electromyography and Inertial Motion Sensing-based Text Entry in AR

Kwon

Shatilov

Lee

et al. 2020

2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)

View full text Add to dashboard Cite

The seamless textual input in Augmented Reality (AR) is very challenging and essential for enabling user-friendly AR applications. Existing approaches such as speech input and vision-based gesture recognition suffer from environmental obstacles and the large default keyboard size, sacrificing the majority of the screen's real estate in AR. In this paper, we propose MyoKey, a system that enables users to effectively and unobtrusively input text in a constrained environment of AR by jointly leveraging surface Electromyography (sEMG) and Inertial Motion Unit (IMU) signals transmitted by wearable sensors on a user's forearm. MyoKey adopts a deep learningbased classifier to infer hand gestures using sEMG. In order to show the feasibility of our approach, we implement a mobile AR application using the Unity application building framework. We present novel interaction and system designs to incorporate information of hand gestures from sEMG and arm motions from IMU to provide seamless text entry solution. We demonstrate the applicability of MyoKey by conducting a series of experiments achieving the accuracy of 0.91 on identifying five gestures in real-time (Inference time: 97.43 ms).

show abstract

Section: B Deep Learning and Mobile Ar Using Semg And Imumentioning

confidence: 99%

MyoKey: Surface Electromyography and Inertial Motion Sensing-based Text Entry in AR

Kwon

Shatilov

Lee

et al. 2020

2020 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)

View full text Add to dashboard Cite

show abstract

“…Assuming the same additive white Gaussian noise (AWGN) channel in transmission for uplink and downlink, the maximum achievable uplink and downlink data rate can be easily derived as [16]:…”

Section: Mes Side Execution Costmentioning

confidence: 99%

“…1. Total offloading scheme (TOS) [16]: TOS is a coarse grained approach. It makes no decision but selects all the components to offload from UE to MES.…”

Section: Experiments Setupmentioning

confidence: 99%

See 1 more Smart Citation

Deep learning-based computation offloading with energy and performance optimization

Gong

Cao

et al. 2020

J Wireless Com Network

View full text Add to dashboard Cite

With the benefit of partially or entirely offloading computations to a nearby server, mobile edge computing gives user equipment (UE) more powerful capability to run computationally intensive applications. However, a critical challenge emerged: how to select the optimal set of components to offload considering the UE performance as well as its battery usage constraints. In this paper, we propose a novel energy and performance efficient deep learning based offloading algorithm. The optimal offloading schemes of components based on remaining energy and its performance can be determined by our proposed algorithm. All of these considerations are modeled as a cost function; then, a deep learning network is trained to compute the solution by which the optimal offloading scheme can be determined. Experimental results show that the proposed method is superior to existing methods in terms of energy and performance constraints.

show abstract

“…In addition, although there exist several offloading methods designed for deep models, they are not focused on the energy problem. For example, Zhang et al [27] design a privacy-preserving offloading model for the security problem; Das et al [28] propose a distributed Stochastic Gradient Descent (SGD) algorithm to accelerate the training speed; Ran et al [29] make a prototype of mobile deep learning application using offloading, in order to implement a realtime Augmented Reality (AR) mobile application.…”

Section: Computation Offloading and Sustainable Learningmentioning

confidence: 99%

Sustainable CNN for Robotic: An Offloading Game in the 3D Vision Computation

Ota

Dong

2019

IEEE Trans. Sustain. Comput.

View full text Add to dashboard Cite

Three-dimensional (3D) scene understanding is of great significance to many robotic applications. With the huge development of the deep learning methods, especially the convolutional neural network (CNN), 3D robotic vision has achieved a satisfactory performance. However, in most scenarios, sustainability becomes a severe problem, and few existing approaches pay enough attention to energy consumption. In this paper, we propose an energy-aware system for sustainable robotic 3D vision. Our contributions mainly include: 1) an effective CNN model for the 3D scene understanding; 2) an offloading strategy to make the deep model more sustainable. First, we design a deep CNN model to analyze the 3D point cloud data. The proposed model contains 92 layers for a state-of-the-art recognition accuracy, which, however, bring a big burden to the computing hardware. Then, we formulate this deep learning computation problem as a non-cooperative game, and adopt a heuristic algorithm to balance the local computing and cloud offloading, in order to obtain an optimal solution, in which both the efficiency and energy-saving are taken into account. Simulations demonstrate that our approach is robust and efficient, and outperforms the state-of-the-art in several related tasks.

show abstract

Delivering Deep Learning to Mobile Devices via Offloading

Cited by 73 publications

References 17 publications

MyoKey: Surface Electromyography and Inertial Motion Sensing-based Text Entry in AR

MyoKey: Surface Electromyography and Inertial Motion Sensing-based Text Entry in AR

Deep learning-based computation offloading with energy and performance optimization

Sustainable CNN for Robotic: An Offloading Game in the 3D Vision Computation

Contact Info

Product

Resources

About