A Comparative Measurement Study of Deep Learning as a Service Framework

被引:20
|
作者
Wu, Yanzhao [1 ]
Liu, Ling [1 ]
Pu, Calton [1 ]
Cao, Wenqi [1 ]
Sahin, Semih [1 ]
Wei, Wenqi [1 ]
Zhang, Qi [2 ]
机构
[1] Georgia Inst Technol, Sch Comp Sci, Atlanta, GA 30332 USA
[2] IBM TJ Watson, Yorktown Hts, NY 10598 USA
基金
美国国家科学基金会;
关键词
Deep learning as a service; big data; deep neural networks; accuracy;
D O I
10.1109/TSC.2019.2928551
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Big data powered Deep Learning (DL) and its applications have blossomed in recent years, fueled by three technological trends: a large amount of digitized data openly accessible, a growing number of DL software frameworks in open source and commercial markets, and a selection of affordable parallel computing hardware devices. However, no single DL framework, to date, dominates in terms of performance and accuracy even for baseline classification tasks on standard datasets, making the selection of a DL framework an overwhelming task. This paper takes a holistic approach to conduct empirical comparison and analysis of four representative DL frameworks with three unique contributions. First, given a selection of CPU-GPU configurations, we show that for a specific DL framework, different configurations of its hyper-parameters may have a significant impact on both performance and accuracy of DL applications. Second, to the best of our knowledge, this study is the first to identify the opportunities for improving the training time performance and the accuracy of DL frameworks by configuring parallel computing libraries and tuning individual and multiple hyper-parameters. Third, we also conduct a comparative measurement study on the resource consumption patterns of four DL frameworks and their performance and accuracy implications, including CPU and memory usage, and their correlations to varying settings of hyper-parameters under different configuration combinations of hardware, parallel computing libraries. We argue that this measurement study provides in-depth empirical comparison and analysis of four representative DL frameworks, and offers practical guidance for service providers to deploying and delivering DL as a Service (DLaaS) and for application developers and DLaaS consumers to select the right DL frameworks for the right DL workloads.
引用
收藏
页码:551 / 566
页数:16
相关论文
共 50 条
  • [1] Comparative Study of Deep Learning Framework in HPC Environments
    Asaadi, Hamidreza
    Chapman, Barbara
    2017 NEW YORK SCIENTIFIC DATA SUMMIT (NYSDS), 2017,
  • [2] MSRDL: Deep learning framework for service recommendation in mashup creation
    Yu, Ting
    Liu, Hailin
    Zhang, Lihua
    Liu, Hongbing
    SCIENTIFIC REPORTS, 2023, 13 (01)
  • [3] MSRDL: Deep learning framework for service recommendation in mashup creation
    Ting Yu
    Hailin Liu
    Lihua Zhang
    Hongbing Liu
    Scientific Reports, 13
  • [4] A Comparative Study on Classification by Deep Learning
    Caliskan, Abdullah
    Badem, Hasan
    Basturk, Alper
    Yuksel, Mehmet Emin
    2016 NATIONAL CONFERENCE ON ELECTRICAL, ELECTRONICS AND BIOMEDICAL ENGINEERING (ELECO), 2016, : 503 - 506
  • [5] Identifying Service Opportunities Based on Outcome-Driven Innovation Framework and Deep Learning: A Case Study of Hotel Service
    Nam, Sunghyun
    Yoon, Sejun
    Raghavan, Nagarajan
    Park, Hyunseok
    SUSTAINABILITY, 2021, 13 (01) : 1 - 25
  • [6] A novel comparative deep learning framework for facial age estimation
    Fatma S. Abousaleh
    Tekoing Lim
    Wen-Huang Cheng
    Neng-Hao Yu
    M. Anwar Hossain
    Mohammed F. Alhamid
    EURASIP Journal on Image and Video Processing, 2016
  • [7] A novel comparative deep learning framework for facial age estimation
    Abousaleh, Fatma S.
    Lim, Tekoing
    Cheng, Wen-Huang
    Yu, Neng-Hao
    Hossain, M. Anwar
    Alhamid, Mohammed F.
    EURASIP JOURNAL ON IMAGE AND VIDEO PROCESSING, 2016,
  • [8] DeepFlow: A Deep Learning Framework For Software-defined Measurement
    Lazaris, Aggelos
    Prasanna, Viktor K.
    CAN'17: PROCEEDINGS OF THE 2017 CLOUD-ASSISTED NETWORKING WORKSHOP, 2017, : 43 - 48
  • [9] MSDF: A Deep Reinforcement Learning Framework for Service Function Chain Migration
    Chen, Ruoyun
    Lu, Hancheng
    Lu, Yujiao
    Liu, Jinxue
    2020 IEEE WIRELESS COMMUNICATIONS AND NETWORKING CONFERENCE (WCNC), 2020,
  • [10] Comparative Study on Crowd Counting with Deep Learning
    Shabbir, Uzair
    Sang, Jun
    Alam, Mohammad S.
    Tan, Jinghan
    Xia, Xiaofeng
    PATTERN RECOGNITION AND TRACKING XXXI, 2020, 11400