数据与计算发展前沿 ›› 2023, Vol. 5 ›› Issue (2): 73-85.

CSTR: 32002.14.jfdc.CN10-1649/TP.2023.02.006

doi: 10.11871/jfdc.issn.2096-742X.2023.02.006

• 专刊：“人工智能&大数据”科研范式变革专刊（上） • 上一篇    下一篇

大数据驱动的海洋人工智能服务平台设计与应用

王凡1(),冯立强1,*(),曹荣强2   

  1. 1.中国科学院海洋研究所，山东 青岛266071
    2.中国科学院计算机网络信息中心，北京100083
  • 收稿日期:2023-03-05 出版日期:2023-04-20 发布日期:2023-04-24
  • 通讯作者: 冯立强
  • 作者简介:王凡，中国科学院海洋研究所，所长、研究员。中国海洋湖沼学会常务副理事长、西北太平洋海洋 环流与气候试验（NPOCE）科学指导委员会主席。长期开展印太交汇区海洋环流与暖池动力学研究。先后主持“973”、国家重点研发计划、国家自然科学基金重点基金和重大基金、中国科学院 战略性先导科技专项等重大项目。针对热带西太平洋环流与暖池的次表层结构与变异、中深层环 流变异等前沿科学问题开展了长期系统研究，取得了重要的科学发现和理论创新，实现深海潜标连续实时观测重大突破，积极开拓“人工智能海洋学”等前沿交叉研究领域。在 Nature、National Science Review、Science Advances 等发表论文 160 余篇；荣获全国优秀科技工作者、中国科学院杰出科技成就奖、山东省自然科学奖一等奖、“海洋工程科技奖一等奖”等。
    本文主要承担的工作为：整体架构设计和海洋科学研究指导。
    WANG Fan, Ph.D., Professor, Director of the Institute of Oceanology, Chinese Academy of Sciences (IOCAS); Vice Chairman, Chinese Society for Oceanology and Limnology; and Chairman, Northwestern Pacific Ocean Circulation and Climate Experiment (NPOCE). He has led National Program on Key Basic Research Project (973 Program), the National Key Research and Development Program, Major and Key Programs of the National Natural Science Foundation of China, the Stra-tegic Priority Research Program of CAS, etc. He carries out re-search on multiscale dynamical processes in the Indo-Pacific convergence area and surrounding regions, and has achieved im-portant scientific discoveries and theory innovations in the three-dimensional circulations in the western Pacific and the subsur-face structure and variabilities in the warm pool. He realized the breakthrough in the realtime transmission of the deep ocean data, and advocated the “artificial intelligence oceanography”. He has published more than 160 scientific papers in journals such as Nature, National Science Review, and Science Advan-ces. He was awarded Outstanding Achievement Prize in Science and Technology of CAS, the first prize of Natural Science of Shandong Province, the first prize of Oceanographic Engin-eering Science and Technology Award, etc. He was also awar-ded National Excellent Scientific and Technological Worker.
    In this paper, he is mainly responsible for the overall frame-work design and research guidance.
    E-mail: fwang@qdio.ac.cn|冯立强，中国科学院海洋研究所，高级工程师，海洋大数据中心副主任，中国海洋学会海洋信息专业委员会委员，青岛市大数据发展促进会专家委员会委员。主要研究方向包括海洋信息技术与应用、海洋科学数据管理与共享、海洋大数据融合与挖掘分析技术研发、海洋大数据与人工智能应用产品研发等。
    本文主要承担工作为：文献调研及平台关键技术方法研究。
    FENG Liqiang, Senior engineer of the Institute of Oceanology, Chinese Academy of Sciences (IOCAS), Deputy Director of Ocean Big Data Center for IOCAS, Member of the Marine Information Professional Committee of the Chinese Society of Oceanography;Member of the Expert Committee of the Qingdao Big Data Development Promotion Association. His main research interests include marine information technology and application, marine science data management and sharing, marine big data fusion and mining analysis technology research and development, and marine big data and artificial intelligence application product research and development.
    In this paper, he is mainly responsible for the literature research and key technology methodology research.
    E-mail: fenglq@qdio.ac.cn
  • 基金资助:
    中国科学院网络安全和信息化专项(CAS-WX2021SF-0108)

Design and Application of Big Data-Driven Ocean Artificial Intelligence Service Platform

WANG Fan1(),FENG Liqiang1,*(),CAO Rongqiang2   

  1. 1. Institute of Oceanology, Chinese Academy of Sciences, Qingdao, Shandong 266071, China
    2. Computer Network Information Center, Chinese Academy of Sciences, Beijing 100083, China
  • Received:2023-03-05 Online:2023-04-20 Published:2023-04-24
  • Contact: FENG Liqiang

RichHTML

26

PDF

139

摘要：

【目的】大数据驱动的海洋人工智能服务平台集成海洋人工智能相关算法、软件工具和数据资源构建支持海洋人工智能研究的科研信息化环境，为科研用户提供“一站式”机器学习即服务。【方法】本文基于海洋领域人工智能研究对数据、软件、算法和定制工作流的实际需求，提出一种大数据驱动的海洋人工智能服务平台框架设计，阐述平台总体架构以及构建海洋人工智能服务平台的关键技术方法，并给出平台支撑实现的海洋人工智能模型研究案例。【结果】大数据驱动的海洋人工智能服务平台提供海洋人工智能模型开发环境，支持数据预处理、特征工程、模型训练、超参数调整、模型评估、模型部署与模型推理等功能，能够帮助海洋领域科研人员搭建海洋人工智能研究工作流，进一步研发面向海洋场景应用的人工智能模型产品。【结论】大数据驱动的海洋人工智能服务平台作为新型科研信息化平台，将促进海洋科学和人工智能的交叉融合，助推研究所人工智能海洋学学科建设，推动海洋大数据与人工智能科研范式变革。

关键词: 科研范式变革, 海洋大数据, 海洋人工智能, 机器学习即服务, 工作流管理, 模型开发s

Abstract:

[Objective] The big data-driven ocean artificial intelligence service platform integrates ocean artificial intelligence-related algorithms, software tools, and data resources to build a scientific research information environment that supports ocean artificial intelligence research. It provides a "one-stop-shop" machine learning as a service for scientific research users. [Methods] Based on the actual needs of data, software, algorithms, and customized workflows in the field of ocean artificial intelligence research, this paper proposes the design of a big data-driven ocean artificial intelligence service platform framework. It describes the overall architecture and the key technical methods for building the platform, and provides a case study of ocean artificial intelligence model research supported by the platform. [Results] The big data-driven ocean AI service platform provides a development environment for ocean AI models, supports functions such as data preprocessing, feature engineering, model training, hyperparameter tuning, model evaluation, model deployment, and model inference. It can assist ocean research scientists in building ocean artificial intelligence research workflows and in developing artificial intelligence model products for ocean scenario applications. [Conclusions] The big data-driven ocean artificial intelligence service platform, as a new type of scientific research information platform, will promote the cross-integration of marine science and artificial intelligence, promote the construction of artificial intelligence oceanography disciplines in IOCAS, and drive the paradigm shift in marine big data and artificial intelligence research.

Key words: transformation of research paradigms, ocean big data, artificial intelligence for oceanography, machine learning as a service, workflow management, model development