Improving the Efficiency of Transformers for Resource-Constrained Devices

被引:12
|
作者
Tabani, Hamid [1 ]
Balasubramaniam, Ajay [1 ]
Marzban, Shabbir [1 ]
Arani, Elahe [1 ]
Zonooz, Bahram [1 ]
机构
[1] NavInfo Europe, Adv Res Lab, Eindhoven, Netherlands
关键词
Deep Learning; Transformers; Clustering; Resource-Constrained Devices;
D O I
10.1109/DSD53832.2021.00074
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
Transformers provide promising accuracy and have become popular and used in various domains such as natural language processing and computer vision. However, due to their massive number of model parameters, memory and computation requirements, they are not suitable for resource-constrained low-power devices. Even with high-performance and specialized devices, the memory bandwidth can become a performance-limiting bottleneck. In this paper, we present a performance analysis of state-of-the-art vision transformers on several devices. We propose to reduce the overall memory footprint and memory transfers by clustering the model parameters. We show that by using only 64 clusters to represent model parameters, it is possible to reduce the data transfer from the main memory by more than 4x, achieve up to 22% speedup and 39% energy savings on mobile devices with less than 0.1% accuracy loss.
引用
收藏
页码:449 / 456
页数:8
相关论文
共 50 条
  • [31] Behavioral fingerprinting to detect ransomware in resource-constrained devices
    Celdran, Alberto Huertas
    Sanchez, Pedro Miguel Sanchez
    von der Assen, Jan
    Shushack, Dennis
    Gomez, angel Luis Perales
    Bovet, Gerome
    Perez, Gregorio Martinez
    Stiller, Burkhard
    COMPUTERS & SECURITY, 2023, 135
  • [32] A web service based agent for resource-constrained devices
    Wang, Xiaodong
    Tao, Ye
    Xu, Xiaowei
    Yu, Zhongqing
    Ding, Xiangqian
    Journal of Information and Computational Science, 2010, 7 (12): : 2443 - 2453
  • [33] XML compression for Web services on resource-constrained devices
    Werner, Christian
    Buschmann, Carsten
    Brandt, Ylva
    Fischer, Stefan
    INTERNATIONAL JOURNAL OF WEB SERVICES RESEARCH, 2008, 5 (03) : 44 - 63
  • [34] Server-guided watermarking for resource-constrained devices
    Jarnikov, Dmitri
    Lourens, J. G.
    Westerveld, Egbert
    ISCE: 2009 IEEE 13TH INTERNATIONAL SYMPOSIUM ON CONSUMER ELECTRONICS, VOLS 1 AND 2, 2009, : 890 - 894
  • [35] Seismic Waveform Inversion Capability on Resource-Constrained Edge Devices
    Manu, Daniel
    Tshakwanda, Petro Mushidi
    Lin, Youzuo
    Jiang, Weiwen
    Yang, Lei
    JOURNAL OF IMAGING, 2022, 8 (12)
  • [36] Hummingbird: Ultra-Lightweight Cryptography for Resource-Constrained Devices
    Engels, Daniel
    Fan, Xinxin
    Gong, Guang
    Hu, Honggang
    Smith, Eric M.
    FINANCIAL CRYPTOGRAPHY AND DATA SECURITY, 2010, 6054 : 3 - +
  • [37] CHAM: A Family of Lightweight Block Ciphers for Resource-Constrained Devices
    Koo, Bonwook
    Roh, Dongyoung
    Kim, Hyeonjin
    Jung, Younghoon
    Lee, Dong-Geon
    Kwon, Daesung
    INFORMATION SECURITY AND CRYPTOLOGY - ICISC 2017, 2018, 10779 : 3 - 25
  • [38] Efficient Online Classification and Tracking on Resource-constrained IoT Devices
    Aftab, Muhammad
    Chau, Sid Chi-Kin
    Shenoy, Prashant
    ACM TRANSACTIONS ON INTERNET OF THINGS, 2020, 1 (03):
  • [39] A Digital Implementation of Extreme Learning Machines for Resource-Constrained Devices
    Ragusa, Edoardo
    Gianoglio, Christian
    Gastaldo, Paolo
    Zunino, Rodolfo
    IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS, 2018, 65 (08) : 1104 - 1108
  • [40] A Hybrid Approach for WebRTC Video Streaming on Resource-Constrained Devices
    Diallo, Bakary
    Ouamri, Abdelaziz
    Keche, Mokhtar
    ELECTRONICS, 2023, 12 (18)