Survey on Network of Distributed Deep Learning Training
-
摘要: 近年来深度学习在图像、语音、自然语言处理等诸多领域得到广泛应用,但随着人们对深度学习的训练速度和数据处理能力的需求不断提升,传统的基于单机的训练过程愈发难以满足要求,分布式的深度学习训练方法成为持续提升算力的有效途径.其中训练过程中节点间网络的通信性能至关重要,直接影响训练性能.分析了分布式深度学习中的性能瓶颈,在此基础上对目前常用的网络性能优化方案进行综述,详细阐述了目前最新的超大规模分布式训练的体系结构、优化方法、训练环境和最有效的优化方法,最后对分布式训练仍然存在的困难进行了总结,对其未来研究方向进行了展望.Abstract: In recent years, deep learning has achieved better results than traditional algorithms in many fields such as image, speech, and natural language processing. People are increasingly demanding training speed and data processing capabilities for deep learning. However, the calculating ability of a single server has a limit and cannot achieve human demands. Distributed deep learning training has become the most effective method to expand deep learning training computing ability. At present, distributed deep learning faces a training bottleneck due to communication problems in the network during the training process which leads the communication network to be the most influential factor. There are currently many network performance optimization researches for distributed deep learning. In this paper, the main performance bottlenecks and optimization schemes are firstly demonstrated. Then the current state-of-art ultra-large-scale distributed training architecture and methods for optimization performance are specifically analyzed. Finally, a comparative summary of each performance optimization scheme and the difficulties still existing in distributed deep learning training are given, and the future research directions are pointed out as well.
-
-
期刊类型引用(15)
1. 王靖,方旭明. Wi-Fi7多链路通感一体化的功率和信道联合智能分配算法. 计算机应用. 2025(02): 563-570 . 百度学术
2. 葛振兴,向帅,田品卓,高阳. 基于深度强化学习的掼蛋扑克博弈求解. 计算机研究与发展. 2024(01): 145-155 . 本站查看
3. Xiaodong Zhuang,Xiangrong Tong. A dynamic algorithm for trust inference based on double DQN in the internet of things. Digital Communications and Networks. 2024(04): 1024-1034 . 必应学术
4. 李迎港,童向荣. 基于知识引导的自适应序列强化学习模型. 模式识别与人工智能. 2023(02): 108-119 . 百度学术
5. 冯景瑜,张静,时翌飞. 物联网中具备终端匿名的加密流量双层过滤方法. 西安邮电大学学报. 2023(02): 72-81 . 百度学术
6. 冯景瑜,李嘉伦,张宝军,韩刚,张文波. 工业互联网中抗APT窃密的主动式零信任模型. 西安电子科技大学学报. 2023(04): 76-88 . 百度学术
7. 丁世飞,杜威,郭丽丽,张健,徐晓. 基于双评论家的多智能体深度确定性策略梯度方法. 计算机研究与发展. 2023(10): 2394-2404 . 本站查看
8. 冯景瑜,王锦康,张宝军,刘宇航. 基于信任过滤的轻量级加密流量异常检测方案. 西安邮电大学学报. 2023(05): 56-66 . 百度学术
9. 徐敏,胡聪,王萍,张翠翠,王鹏. 基于强化学习的Ceph文件系统的性能优化. 微型电脑应用. 2022(03): 83-86 . 百度学术
10. 冯景瑜,于婷婷,王梓莹,张文波,韩刚,黄文华. 电力物联场景下抗失陷终端威胁的边缘零信任模型. 计算机研究与发展. 2022(05): 1120-1132 . 本站查看
11. 王鑫,赵清杰,于重重,张长春,陈涌泉. 多节点探测器软着陆的路径规划方法. 宇航学报. 2022(03): 366-373 . 百度学术
12. 张文璐,霍子龙,赵西雨,崔琪楣,陶小峰. 面向智能工厂多机器人定位的无线分布式协同决策. 无线电通信技术. 2022(04): 718-727 . 百度学术
13. 王岩,童向荣. 基于tri-training和极限学习机的跨领域信任预测. 计算机研究与发展. 2022(09): 2015-2026 . 本站查看
14. 聂雷,刘博,李鹏,何亨. 基于多智能体Q学习的异构车载网络选择方法. 计算机工程与科学. 2021(05): 836-844 . 百度学术
15. 洪志理,赖俊,曹雷,陈希亮. 融合用户兴趣建模的智能推荐算法研究. 信息技术与网络安全. 2021(11): 37-48 . 百度学术
其他类型引用(15)
计量
- 文章访问数: 2262
- HTML全文浏览量: 22
- PDF下载量: 1808
- 被引次数: 30