网络智能化中的AI工程化技术方案

doi:10.11959/j.issn.1000-0801.2022016

摘要/Abstract

摘要：

网络智能化是通信行业借助 AI 技术，对外增强网络赋能能力，对内实现降本增效的重要举措。从AI工程化的视角系统分析网络智能化应用落地的难点，提出了包括数据采集处理、训练计算资源的管理与任务调度、推理部署优化在内的面向生产环境的AI工程化技术方案，探讨网络智能化生态发展的策略。

关键词: 网络智能化, 人工智能, 云原生, 模型压缩, 推理服务

Abstract:

Depending on AI technology, network intelligence is becoming an important initiative for communication industry to enhance network empowerment externally, and to achieve cost reduction and efficiency internally.The difficulties implementing network intelligence applications from the perspective of AI engineering were analyzed.The industrial grade AI engineering technical solutions were proposed, including data collection and processing, computing resources management and task scheduling, and inference deployment optimization.The strategies of network intelligence’s ecosystem development were studied.

Key words: network intelligence, artificial intelligence, cloud native, model compression, inference service

中图分类号:

朱明伟. 网络智能化中的AI工程化技术方案[J]. 电信科学, 2022, 38(2): 157-165.

Mingwei ZHU. AI engineering technology solutions in network intelligence[J]. Telecommunications Science, 2022, 38(2): 157-165.

图/表 5

参考文献 29

[1]	JIA W L , WANG H , CHEN M H ,et al. Pushing the limit of molecular dynamics with ab initio accuracy to 100 million atoms with machine learning[J]. arXiv:2005.00223, 2020.
[2]	JUMPER J , EVANS R , PRITZEL A ,et al. Highly accurate protein structure prediction with AlphaFold[J]. Nature, 2021,596(7873): 583-589.
[3]	欧阳晔, 王立磊, 杨爱东 ,等. 通信人工智能的下一个十年[J]. 电信科学, 2021,37(3): 1-36.
	OUYANG Y , WANG L L , YANG A D ,et al. Next decade of telecommunications artificial intelligence[J]. Telecommunica-tions Science, 2021,37(3): 1-36.
[4]	华为技术有限公司. 华为自动驾驶网络解决方案白皮书[R]. 2020.
	Huawei Technologies Co.,Ltd.. Huawei's white paper on auto-nomous driving network solutions[R]. 2020.
[5]	中兴通讯股份有限公司. 中兴自主进化网络白皮书[R]. 2020.
	ZTE Technology Co.,Ltd.. ZTE’s white paper on autonomous evolution network[R]. 2020.
[6]	邓超, 王斌, 朱琳 ,等. 人工智能在电信运营中的典型应用实践[J]. 信息通信技术与政策, 2019(7): 34-38.
	DENG C , WANG B , ZHU L ,et al. Typical applications of ar-tificial intelligence in telecom operation[J]. Information and Communications Technology and Policy, 2019(7): 34-38.
[7]	冯俊兰 . 5G自身智能化及赋能智能产业之路[J]. 电信工程技术与标准化, 2020,33(1): 1-8.
	FENG J L . Intelligent 5G network and 5G+AI applications[J]. Tel-ecom Engineering Technics and Standardization, 2020,33(1): 1-8.
[8]	程强, 刘姿杉 . 数据驱动的智能电信网络[J]. 中兴通讯技术, 2020,26(5): 53-56.
	CHENG Q , LIU Z S . Data empowered intelligent communica-tion networks[J]. ZTE Technology Journal, 2020,26(5): 53-56.
[9]	RATNER A , BACH S H , EHRENBERG H ,et al. Snorkel:rapid training data creation with weak supervision[J]. The VLDB Journal:Very Large Data Bases:a Publication of the VLDB Endowment, 2020,29(2): 709-730.
[10]	BRECK E , CAI S Q , NIELSEN E ,et al. The ML test score:a rubric for ML production readiness and technical debt reduction[C]// Proceedings of 2017 IEEE International Conference on Big Data (Big Data). Piscataway:IEEE Press, 2017: 1123-1132.
[11]	CNCF. Cloud Native Interactive Landscape[EB]. 2021.
[12]	Kubeflow. Kubeflow Overview[EB]. 2021.
[13]	Github. Volcano[EB]. 2021.
[14]	Kubernetes. Scheduling Framework[EB]. 2021.
[15]	BROWN T B , MANN B , RYDER N ,et al. Language models are few-shot learners[J]. arXiv:2005.14165, 2020.
[16]	Horovod. Elastic Horovod[EB]. 2021.
[17]	GIBIANSKY A . Bringing HPC techniques to deep learning[EB]. 2021.
[18]	HOWARD A G , ZHU M L , CHEN B ,et al. MobileNets:efficient convolutional neural networks for mobile vision applications[EB]. 2017:arXiv:1704.04861, 2017.
[19]	HAN S , MAO H Z , DALLY W . Deep compression:compressing deep neural network with pruning,trained quantization and huffman coding[J]. ICLR. 2015.
[20]	INTEL Accelerate lower numerical precision inference with Intel? deep learning boost[EB]. 2021.
[21]	HINTON G , VINYALS O , DEAN J . Distilling the knowledge in a neural network[J]. Computer Science, 2015,14(7): 38-39.
[22]	LI M Z , LIU Y , LIU X Y ,et al. The deep learning compiler:a comprehensive survey[J]. IEEE Transactions on Parallel and Distributed Systems, 2021,32(3): 708-727.
[23]	Github KFServing:predict on an InferenceService with TensorFlow model[EB]. 2021.
[24]	TensorFlow. TFServing：train and serve a TensorFlow model with TensorFlow serving[EB]. 2021.
[25]	Google Cloud. MLOps：continuous delivery and automated pipelines in machine learning[EB]. 2021.
[26]	Rancher. K3s[EB]. 2021.
[27]	GitHub. KubeEdge[EB]. 2021.
[28]	Gitee. OpenYurt[EB]. 2021.
[29]	刘腾飞, 李奥 . Acumos:一种人工智能开放平台[J]. 邮电设计技术, 2018(12): 46-50.
	LIU T F , LI A . Acumos—an artificial intelligence open plat-form[J]. Designing Techniques of Posts and Telecommunica-tions, 2018(12): 46-50.