摘要
[目的/意义]探讨迁移学习技术研究现状及在信息资源开发及服务中的应用前景。[方法/过程]在总结和阐述迁移学习研究价值和研究进展的基础上,构建层次化的信息资源开发及服务结构体系,探讨迁移学习和信息资源开发及服务的融合发展,展望基于迁移学习的信息资源开发及服务研究趋势。[结果/结论]迁移学习克服了传统机器学习方法对“大数据+大模型+大计算”的依赖,打破了“从零学习”的固有范式,更适合解决小样本场景和个性化领域中的数据挖掘和分析问题,为大规模复杂环境下信息资源开发及服务探索提供一种新的技术手段和实践思路。
[Purpose/significance]This paper discusses the research status of transfer learning technology and its application prospect in information resource development and service.[Method/process]On the basis of expounding and summarizing the research value and research progress of transfer learning,this paper constructs a hierarchical information resource development and service structure system,probes into the integrated development of transfer learning and information resource development and service,and predicts the research trend of information resource development and service based on transfer learning.[Result/conclusion]Transfer learning overcomes the dependence of traditional machine learning method on big data,model and calculation,breaks the inherent ab initio pattern,which is more suitable to solve problems of mining and analysis for small sample and personized data.It provides a new idea for information resources development and service in the large-scale complicated environment.
出处
《情报理论与实践》
CSSCI
北大核心
2021年第7期145-151,共7页
Information Studies:Theory & Application
关键词
深度学习
迁移学习
数据挖掘
融合发展
小样本学习
信息资源开发及服务
deep learning
transfer learning
data mining
fusion development
few-shot learning
information resource development and service