摘要
人工智能与机器识别技术的发展加快了社会的进步,传统机器学习的方法并不适用于所有的环境,这就要求目标识别算法能够在半监督或无监督情况下进行训练。本文提出基于结构化联合分布适配的无监督大样本跨领域目标识别算法模型。为实现无监督少样本条件下的跨领域目标识别提供新的思路与方法。
The development of artificial intelligence and machine recognition technology has accelerated the progress of society.Traditional machine learning methods are not applicable to all environments,which requires that target recognition algorithms can be trained under semi-supervised or unsupervised conditions.This paper presents an unsupervised large sample cross-domain target recognition algorithm model based on structured joint distributed adaptation.It provides new ideas and methods for cross-domain target recognition under the condition of unsupervised and few samples.
作者
周丽丽
杜寅甫
ZHOU Li-li;DU Yin-fu(Institute of Intelligent Manufacturing,Heilongjiang Academy of Sciences,Harbin 150090 China;Heilongjiang Academy of Sciences High Technology Research Institute,Harbin 150020 China)
出处
《自动化技术与应用》
2019年第11期168-171,共4页
Techniques of Automation and Applications
关键词
目标监测与识别
无监督学习
跨领域识别
target monitoring and recognition
unsupervised learning
cross-domain recognition
