电信科学 ›› 2023, Vol. 39 ›› Issue (6): 44-51.doi: 10.11959/j.issn.1000-0801.2023128

• 专题:多模态网络 • 上一篇    下一篇

面向分布式机器学习的网络模态创新

郭泽华1, 朱昊文1, 徐同文1,2   

  1. 1 北京理工大学自动化学院,北京 100081
    2 延安大学物理与电子信息学院,陕西 延安 716099
  • 修回日期:2023-06-09 出版日期:2023-06-20 发布日期:2023-06-01
  • 作者简介:郭泽华(1985- ),男,博士,北京理工大学自动化学院特别研究员,主要研究方向为可编程网络、机器学习和网络安全
    朱昊文(1992- ),男,北京理工大学自动化学院博士生,主要研究方向为分布式机器学习与可编程网络
    徐同文(1990- ),男,北京理工大学自动化学院博士生、延安大学物理与电子信息学院讲师,主要研究方向为计算机网络
  • 基金资助:
    国家自然科学基金资助项目(62002019);CCF-之江实验室联合创新基金(K2022QA0AB02);嵩山实验室预研项目(YYJC022022009)

Network modal innovation for distributed machine learning

Zehua GUO1, Haowen ZHU1, Tongwen XU1,2   

  1. 1 School of Automation, Beijing Institute of Technology, Beijing 100081, China
    2 School of Physics and Electronic Information, Yan’an University, Yan’an 716099, China
  • Revised:2023-06-09 Online:2023-06-20 Published:2023-06-01
  • Supported by:
    The National Natural Science Foundation of China(62002019);CCF-Zhijiang Laboratory Joint Innovation Fund(K2022QA0AB02);Songshan Laboratory Pre-research Project(YYJC022022009)

摘要:

分布式机器学习作为人工智能的主流计算架构,目前仍然存在数据性能传输不高、模型训练速度慢等缺陷,传统的网络模态无法满足分布式机器学习场景的通信语义,继而无法解决这些缺陷以进一步提升模型训练性能。采用多模态网络技术,基于应用特点设计了面向分布式机器学习场景的新型网络模态及其运行逻辑,为多模态网络技术在垂直行业的应用提供了借鉴意义。

关键词: 多模态网络, 分布式机器学习, 模型训练, 人工智能

Abstract:

Distributed machine learning, as a popular computing architecture for artificial intelligence, still faces challenges of slow model training and poor data performance transmission.Traditional network modalities were un able to meet the communication needs of distributed machine learning scenarios, hindering the improvement of model training performance.New network modalities and operation logic for distributed machine learning scenarios using multimodal network technology were proposed.This approach was designed based on application characteristics and provides implications for the use of multimodal network technology in various industries.

Key words: multimodal network, distributed machine learning, model training, artificial intelligence

中图分类号: 

No Suggested Reading articles found!