Institut for Elektro- og Computerteknologi

Aarhus Universitets segl
Du er her: Institut for Elektro- og Computerteknologi » Om instituttet » Kontakt » Qi Zhang » Publikationer » Receptive Field-based Segmentation for Distributed CNN In...

Om instituttet

Qi Zhang

Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing

Publikation: Bidrag til bog/antologi/rapport/proceeding › Konferencebidrag i proceedings › Forskning › peer review

Standard

Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing. / Li, Nan; Iosifidis, Alexandros; Zhang, Qi.
ICC 2022 - IEEE International Conference on Communications. IEEE, 2022. s. 4281-4286.

Publikation: Bidrag til bog/antologi/rapport/proceeding › Konferencebidrag i proceedings › Forskning › peer review

Harvard

Li, N, Iosifidis, A & Zhang, Q 2022, Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing. i ICC 2022 - IEEE International Conference on Communications. IEEE, s. 4281-4286, IEEE International Conference on Communications, Seoul, Sydkorea, 16/05/2022. https://doi.org/10.1109/ICC45855.2022.9838547

APA

Li, N., Iosifidis, A., & Zhang, Q. (2022). Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing. I ICC 2022 - IEEE International Conference on Communications (s. 4281-4286). IEEE. https://doi.org/10.1109/ICC45855.2022.9838547

CBE

Li N, Iosifidis A, Zhang Q. 2022. Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing. I ICC 2022 - IEEE International Conference on Communications. IEEE. s. 4281-4286. https://doi.org/10.1109/ICC45855.2022.9838547

MLA

Li, Nan, Alexandros Iosifidis, og Qi Zhang "Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing". ICC 2022 - IEEE International Conference on Communications. IEEE. 2022, 4281-4286. https://doi.org/10.1109/ICC45855.2022.9838547

Vancouver

Li N, Iosifidis A, Zhang Q. Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing. I ICC 2022 - IEEE International Conference on Communications. IEEE. 2022. s. 4281-4286 doi: 10.1109/ICC45855.2022.9838547

Author

Li, Nan ; Iosifidis, Alexandros ; Zhang, Qi. / Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing. ICC 2022 - IEEE International Conference on Communications. IEEE, 2022. s. 4281-4286

Bibtex

@inproceedings{942da6c88df54d06a738ca3bf270a6d8,
title = "Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing",
abstract = "This paper studies inference acceleration using distributed convolutional neural networks (CNNs) in collaborative edge computing network. To avoid inference accuracy loss in inference task partitioning, we propose receptive field-based segmentation (RFS). To reduce the computation time and communication overhead, we propose a novel collaborative edge computing using fused-layer parallelization to partition a CNN model into multiple blocks of convolutional layers. In this scheme, the collaborative edge servers (ESs) only need to exchange small fraction of the sub-outputs after computing each fused block. In addition, to find the optimal solution of partitioning a CNN model into multiple blocks, we use dynamic programming, named as dynamic programming for fused-layer parallelization (DPFP). The experimental results show that DPFP can accelerate inference of VGG-16 up to 73% compared with the pre-trained model, which outperforms the existing work MoDNN in all tested scenarios. Moreover, we evaluate the service reliability of DPFP under time-variant channel, which shows that DPFP is an effective solution to ensure high service reliability with strict service deadline.",
keywords = "Distributed CNNs, Edge computing, Receptive field, Service reliability, Time-critical IoT",
author = "Nan Li and Alexandros Iosifidis and Qi Zhang",
note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; IEEE International Conference on Communications ; Conference date: 16-05-2022 Through 20-05-2022",
year = "2022",
doi = "10.1109/ICC45855.2022.9838547",
language = "English",
isbn = "978-1-5386-8348-4 ",
pages = "4281--4286",
booktitle = "ICC 2022 - IEEE International Conference on Communications",
publisher = "IEEE",
url = "https://icc2022.ieee-icc.org/",

}

RIS

TY - GEN

T1 - Receptive Field-based Segmentation for Distributed CNN Inference Acceleration in Collaborative Edge Computing

AU - Li, Nan

AU - Iosifidis, Alexandros

AU - Zhang, Qi

N1 - Publisher Copyright: © 2022 IEEE.

PY - 2022

Y1 - 2022

N2 - This paper studies inference acceleration using distributed convolutional neural networks (CNNs) in collaborative edge computing network. To avoid inference accuracy loss in inference task partitioning, we propose receptive field-based segmentation (RFS). To reduce the computation time and communication overhead, we propose a novel collaborative edge computing using fused-layer parallelization to partition a CNN model into multiple blocks of convolutional layers. In this scheme, the collaborative edge servers (ESs) only need to exchange small fraction of the sub-outputs after computing each fused block. In addition, to find the optimal solution of partitioning a CNN model into multiple blocks, we use dynamic programming, named as dynamic programming for fused-layer parallelization (DPFP). The experimental results show that DPFP can accelerate inference of VGG-16 up to 73% compared with the pre-trained model, which outperforms the existing work MoDNN in all tested scenarios. Moreover, we evaluate the service reliability of DPFP under time-variant channel, which shows that DPFP is an effective solution to ensure high service reliability with strict service deadline.

AB - This paper studies inference acceleration using distributed convolutional neural networks (CNNs) in collaborative edge computing network. To avoid inference accuracy loss in inference task partitioning, we propose receptive field-based segmentation (RFS). To reduce the computation time and communication overhead, we propose a novel collaborative edge computing using fused-layer parallelization to partition a CNN model into multiple blocks of convolutional layers. In this scheme, the collaborative edge servers (ESs) only need to exchange small fraction of the sub-outputs after computing each fused block. In addition, to find the optimal solution of partitioning a CNN model into multiple blocks, we use dynamic programming, named as dynamic programming for fused-layer parallelization (DPFP). The experimental results show that DPFP can accelerate inference of VGG-16 up to 73% compared with the pre-trained model, which outperforms the existing work MoDNN in all tested scenarios. Moreover, we evaluate the service reliability of DPFP under time-variant channel, which shows that DPFP is an effective solution to ensure high service reliability with strict service deadline.

KW - Distributed CNNs

KW - Edge computing

KW - Receptive field

KW - Service reliability

KW - Time-critical IoT

U2 - 10.1109/ICC45855.2022.9838547

DO - 10.1109/ICC45855.2022.9838547

M3 - Article in proceedings

SN - 978-1-5386-8348-4

SP - 4281

EP - 4286

BT - ICC 2022 - IEEE International Conference on Communications

PB - IEEE

T2 - IEEE International Conference on Communications

Y2 - 16 May 2022 through 20 May 2022

ER -