基于多视图未标记数据的机器学习

Research and Advances on Machine Learning of Multi-view Unlabeled Data

半监督学习和主动学习,与传统的监督学习不同,能同时在少量的已标记数据和大量的未标记数据上进行学习,从而提高性能。半监督学习和主动学习,最初是建立在单视图数据上的,但最近的研究表明对多视图数据,它们也能产生很好效果。本文综述多视图数据半监督学习和主动学习基本思想、常用算法和最新研究进展,并指出需进一步研究的几个问题。

Different from traditional supervised-learning,semi-supervised learning and active learning can use unlabeled data together with labeled data to improve the performance.Although these techniques are initially developed for date with a single view,recent studies show that for multi-view data,semi-supervised and active learning can work well.This paper presents a survey of the main problems and the state-of-art multi-view learning algorithms.Main difficulties and questions which will be solved in the future are also shown.