A Dynamic Deep Neural Network Design for Efficient Workload Allocation in Edge Computing

Lo, Chi; Su, Yu-Yi; Lee, Chun-Yi; Chang, Shih-Chieh

doi:10.1109/iccd.2017.49

Cited by 40 publications

(26 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In DeepIns, edge devices are responsible for data collection, the edge server acts as the first exit point and the cloud data center acts as the second exit point. Then Lo et al [90] proposes adding an authentic operation (AO) unit to the basic BranchyNet model. The AO unit determines whether an input has to be transferred to the edge server or cloud data center for further execution by setting different threshold criteria of confidence level for different DNN model output classes.…”

Section: Enabling Technologiesmentioning

confidence: 99%

Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing

et al. 2019

View full text Add to dashboard Cite

With the breakthroughs in deep learning, the recent years have witnessed a booming of artificial intelligence (AI) applications and services, spanning from personal assistant to recommendation systems to video/audio surveillance. More recently, with the proliferation of mobile computing and Internet-of-Things (IoT), billions of mobile and IoT devices are connected to the Internet, generating zillions Bytes of data at the network edge. Driving by this trend, there is an urgent need to push the AI frontiers to the network edge so as to fully unleash the potential of the edge big data. To meet this demand, edge computing, an emerging paradigm that pushes computing tasks and services from the network core to the network edge, has been widely recognized as a promising solution.The resulted new inter-discipline, edge AI or edge intelligence, is beginning to receive a tremendous amount of interest. However, research on edge intelligence is still in its infancy stage, and a dedicated venue for exchanging the recent advances of edge intelligence is highly desired by both the computer system and artificial intelligence communities. To this end, we conduct a comprehensive survey of the recent research efforts on edge intelligence. Specifically, we first review the background and motivation for artificial intelligence running at the network edge. We then provide an overview of the overarching architectures, frameworks and emerging key technologies for deep learning model towards training/inference at the network edge. Finally, we discuss future research opportunities on edge intelligence. We believe that this survey will elicit escalating attentions, stimulate fruitful discussions and inspire further research ideas on edge intelligence.

show abstract

Section: Enabling Technologiesmentioning

confidence: 99%

Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing

et al. 2019

View full text Add to dashboard Cite

show abstract

“…This is because the data dependency exists between each pair of DNN layers. Lo et al [16] presented a dynamic DNN design technique to manage the workload transmission under the same accuracy requirement in edge computing. They utilized the dynamic network structure and authentic operation (AO) unit to enhance DNNs, which had a better performance in terms of reducing the amount of workload transmission while achieving the required accuracy.…”

Section: Related Workmentioning

confidence: 99%

Energy-Efficient Offloading for DNN-Based Smart IoT Systems in Cloud-Edge Environments

Chen

Zhang

Lin

et al. 2022

IEEE Trans. Parallel Distrib. Syst.

112

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) have become an essential and important supporting technology for smart Internet-of-Things (IoT) systems. Due to the high computational costs of large-scale DNNs, it might be infeasible to directly deploy them in energy-constrained IoT devices. Through offloading computation-intensive tasks to the cloud or edges, the computation offloading technology offers a feasible solution to execute DNNs. However, energy-efficient offloading for DNN based smart IoT systems with deadline constraints in the cloud-edge environments is still an open challenge. To address this challenge, we first design a new system energy consumption model, which takes into account the runtime, switching, and computing energy consumption of all participating servers (from both the cloud and edge) and IoT devices. Next, a novel energy-efficient offloading strategy based on a Self-adaptive Particle Swarm Optimization algorithm using the Genetic Algorithm operators (SPSO-GA) is proposed. This new strategy can efficiently make offloading decisions for DNN layers with layer partition operations, which can lessen the encoding dimension and improve the execution time of SPSO-GA. Simulation results demonstrate that the proposed strategy can significantly reduce energy consumption compared to other classic methods.

show abstract

“…Model compression, such as pruning [46], data quantization [47] and knowledge distillation [48], can minimize the model complexity to relieve the pressure of the end devices. Model early-exit [49], [50], model selection [51], model partition [52], [53] and input filtering [54], [55] can realize inference acceleration, further speeding up the deployment on edge devices with the limited memory.…”

Section: Model Establishmentmentioning

confidence: 99%

Synergy of Edge Intelligence and Blockchain: A Comprehensive Survey

Wang¹,

Ren²,

Qiu³

et al. 2021

Preprint

View full text Add to dashboard Cite

Edge intelligence (EI), as an emerging technology, has been attracting significant attention. It pushes the frontier of artificial intelligence (AI) from the cloud to the network edge, aiming to embrace and support the next-generation communications while unleashing AI services. However, it faces challenges in its decentralized management and security, which limit its capabilities to support services with numerous requirements. On the other side, blockchain (BC), as a promising decentralized technology, is beneficial to tackle the above issues. However, there exist some technical challenges for BC, such as transaction capacity, scalability, and fault tolerance. Motivated by significant current interest around EI and BC, this survey examines whether the synergy of EI and BC can make a powerful network with combined functionalities of both cutting-edge technologies. Accordingly, we develop EI-chain and Chain-intelligence to realize reliable computing-power management, data administration, and model optimization at the edges, while improving the functions of BC by leveraging complementary characteristics of EI and BC, further making up for their own limitations. In this survey, a wide spectrum of literature is carefully reviewed to enable the EI-chain and Chain-intelligence. Moreover, we cover the technological aspects of EI-chain and Chain-intelligence: overview, motivations, and frameworks. Finally, some challenges and future directions are explored. We believe this survey will provide developers and researchers a comprehensive view on the synergy of EI and BC, while accelerating the design of a powerful integrated network.

show abstract

A Dynamic Deep Neural Network Design for Efficient Workload Allocation in Edge Computing

Cited by 40 publications

References 12 publications

Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing

Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing

Energy-Efficient Offloading for DNN-Based Smart IoT Systems in Cloud-Edge Environments

Synergy of Edge Intelligence and Blockchain: A Comprehensive Survey

Contact Info

Product

Resources

About