面向认知的多源数据学习理论和算法研究进展
发布时间：2018-08-24 12:39
【摘要】：多源数据学习在大数据时代具有极其重要的意义.目前,多源数据学习算法研究远远超前于多源数据学习理论研究,经典的机器学习理论难以应用于多源数据学习,更难以提供多源数据学习算法在实际应用中的理论保障.从学习的最终目的是知识这一认知切入点出发,对人类学习的认知机理、机器学习的三大经典理论(计算学习理论、统计学习理论和概率图理论)以及多源数据学习算法设计这3个方面的研究进展进行总结,最后给出未来研究方向的思考.
[Abstract]:Multi-source data learning is of great significance in big data's time. At present, the research of multi-source data learning algorithm is far ahead of the theory of multi-source data learning, the classical machine learning theory is difficult to be applied to multi-source data learning, and it is difficult to provide the theoretical guarantee of multi-source data learning algorithm in practical application. Starting from the point of view that the ultimate aim of learning is knowledge, the cognitive mechanism of human learning, the three classical theories of machine learning (computational learning theory), Statistical learning theory and probability graph theory) and the design of multi-source data learning algorithm are summarized. Finally, the future research direction is given.
【作者单位】： 天津大学计算机科学与技术学院;北京交通大学交通数据分析与挖掘北京市重点实验室;脑与认知科学国家重点实验室(中国科学院心理研究所);中国科学院大学心理学系;南京大学计算机软件新技术国家重点实验室;
【分类号】：TP181
本文编号：2200844
[Abstract]:Multi-source data learning is of great significance in big data's time. At present, the research of multi-source data learning algorithm is far ahead of the theory of multi-source data learning, the classical machine learning theory is difficult to be applied to multi-source data learning, and it is difficult to provide the theoretical guarantee of multi-source data learning algorithm in practical application. Starting from the point of view that the ultimate aim of learning is knowledge, the cognitive mechanism of human learning, the three classical theories of machine learning (computational learning theory), Statistical learning theory and probability graph theory) and the design of multi-source data learning algorithm are summarized. Finally, the future research direction is given.
【作者单位】： 天津大学计算机科学与技术学院;北京交通大学交通数据分析与挖掘北京市重点实验室;脑与认知科学国家重点实验室(中国科学院心理研究所);中国科学院大学心理学系;南京大学计算机软件新技术国家重点实验室;
【分类号】：TP181
【相似文献】
相关博士学位论文 前2条
1 何斌;基于可拓逻辑的机器学习理论与方法[D];华南理工大学;2005年
2 张利军;大规模机器学习理论研究与应用[D];浙江大学;2012年
相关硕士学位论文 前1条
1 裴岩;机器学习理论研究及其在车载导航系统中的应用[D];东北大学 ;2009年
，本文编号：2200844
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2200844.html
