基于分歧的半监督学习

传统监督学习通常需使用大量有标记的数据样本作为训练例,而在很多现实问题中,人们虽能容易地获得大批数据样本,但为数据 提供标记却需耗费很多人力物力.那么,在仅有少量有标记数据时,可否通过对大量未标记数据进行利用来提升学习性能呢？为此,半监督学习 成为近十多年来机器学习的一大研究热点.基于分歧的半监督学习是该领域的主流范型之一,它通过使用多个学习器来对未标记数据进行利用, 而学习器间的"分歧"对学习成效至关重要.本文将综述简介这方面的一些研究进展.     

基于半监督回归的选择性集成算法

为了提高小数据量的有标记样本问题中学习器的性能,结合半监督学习和选择性集成学习,提出了基于半监督回归的选择性集成算法SSRES。算法基于半监督学习的基本思想,同时使用有标记样本和未标记样本训练学习器从而减少对有标记样本的需求,使用选择性集成算法GRES对不同学习器进行适当的选择,并将选择的结果结合提高学习器的泛化能力。实验结果表明,在小数据量的有标记样本问题中,该算法能够有效地提高学习器的性能。     

Multiple-View Multiple-Learner Semi-Supervised Learning

Some recent successful semi-supervised learning methods construct more than one learner from both labeled and unlabeled data for inductive learning. This paper proposes a novel multiple-view multiple-learner (MVML) framework for semi-supervised learning, which differs from previous methods in possession of both multiple views and multiple learners. This method adopts a co-training styled learning paradigm in enlarging labeled data from a much larger set of unlabeled data. To the best of our knowledge it is the first attempt to combine the advantages of multiple-view learning and ensemble learning for semi-supervised learning. The use of multiple views is promising to promote performance compared with single-view learning because information is more effectively exploited. At the same time, as an ensemble of classifiers is learned from each view, predictions with higher accuracies can be obtained than solely adopting one classifier from the same view. Experiments on different applications involving both multiple-view and single-view data sets show encouraging results of the proposed MVML method.     

Instance selection method for improving graph-based semi-supervised learning

Graph-based semi-supervised learning is an important semi-supervised learning paradigm. Although graph-based semi-supervised learning methods have been shown to be helpful in various situations, they may adversely affect performance when using unlabeled data. In this paper, we propose a new graph-based semi-supervised learning method based on instance selection in order to reduce the chances of performance degeneration. Our basic idea is that given a set of unlabeled instances, it is not the best approach to exploit all the unlabeled instances; instead, we should exploit the unlabeled instances that are highly likely to help improve the performance, while not taking into account the ones with high risk. We develop both transductive and inductive variants of our method. Experiments on a broad range of data sets show that the chances of performance degeneration of our proposed method are much smaller than those of many state-of-the-art graph-based semi-supervised learning methods.     

Exploiting unlabeled data to enhance ensemble diversity

Ensemble learning learns from the training data by generating an ensemble of multiple base learners. It is well-known that to construct a good ensemble with strong generalization ability, the base learners are deemed to be accurate as well as diverse. In this paper, unlabeled data is exploited to facilitate ensemble learning by helping augment the diversity among the base learners. Specifically, a semi-supervised ensemble method named udeed, i.e. Unlabeled Data to Enhance Ensemble Diversity, is proposed. In contrast to existing semi-supervised ensemble methods which utilize unlabeled data by estimating error-prone pseudo-labels on them to enlarge the labeled data to improve base learners’ accuracies, udeed works by maximizing accuracies of base learners on labeled data while maximizing diversity among them on unlabeled data. Extensive experiments on 20 regular-scale and five large-scale data sets are conducted under the setting of either few or abundant labeled data. Experimental results show that udeed can effectively utilize unlabeled data for ensemble learning via diversity augmentation, and is highly competitive to well-established semi-supervised ensemble methods.     

基于Tri—Training算法的数据编辑技术

Tri-Training是一种半监督学习算法,在少量标记数据下,通过三个不同的分类器,从未标记样本中采样并标记新的训练数据,作为各分类器训练数据的有效补充。但由于错误标记样本的存在,引入了噪音数据,降低了分类的性能。论文在Tri—Training算法中分别采用DE-KNN,DE-BKNN和DE-NED三种数据编辑技术,识别移除误标记的数据。通过对六组UCI数据集的实验,分析结果表明,编辑技术的引入是有效的,三种方法的使用在一定程度上提升了Tri-Training算法的分类性能,尤其是DE-NED方法更为显著。     

基于多学习器协同训练模型的人体行为识别方法

人体行为识别是计算机视觉研究的热点问题,现有的行为识别方法都是基于监督学习框架.为了取得较好的识别效果,通常需要大量的有标记样本来建模.然而,获取有标记样本是一个费时又费力的工作.为了解决这个问题,对半监督学习中的协同训练算法进行改进,提出了一种基于多学习器协同训练模型的人体行为识别方法.这是一种基于半监督学习框架的识别算法.该方法首先通过基于Q统计量的学习器差异性度量选择算法来挑取出协同训练中基学习器集,在协同训练过程中,这些基学习器集对未标记样本进行标记;然后,采用了基于分类器成员委员会的标记近邻置信度计算公式来评估未标记样本的置信度,选取一定比例置信度较高的未标记样本加入到已标记的训练样本集并更新学习器来提升模型的泛化能力.为了评估算法的有效性,采用混合特征来表征人体行为,从而可以快速完成识别过程.实验结果表明,所提出的基于半监督学习的行为识别系统可以有效地辨识视频中的人体动作.     

SemiBoost: Boosting for Semi-Supervised Learning

Semi-supervised learning has attracted a significant amount of attention in pattern recognition and machine learning. Most previous studies have focused on designing special algorithms to effectively exploit the unlabeled data in conjunction with labeled data. Our goal is to improve the classification accuracy of any given supervised learning algorithm by using the available unlabeled examples. We call this as the Semi-supervised improvement problem, to distinguish the proposed approach from the existing approaches. We design a metasemi-supervised learning algorithm that wraps around the underlying supervised algorithm and improves its performance using unlabeled data. This problem is particularly important when we need to train a supervised learning algorithm with a limited number of labeled examples and a multitude of unlabeled examples. We present a boosting framework for semi-supervised learning, termed as SemiBoost. The key advantages of the proposed semi-supervised learning approach are: 1) performance improvement of any supervised learning algorithm with a multitude of unlabeled data, 2) efficient computation by the iterative boosting algorithm, and 3) exploiting both manifold and cluster assumption in training classification models. An empirical study on 16 different data sets and text categorization demonstrates that the proposed framework improves the performance of several commonly used supervised learning algorithms, given a large number of unlabeled examples. We also show that the performance of the proposed algorithm, SemiBoost, is comparable to the state-of-the-art semi-supervised learning algorithms.     

Combining Committee-Based Semi-Supervised Learning and Active Learning

Many data mining applications have a large amount of data but labeling data is usually difficult, expensive, or time consuming, as it requires human experts for annotation. Semi-supervised learning addresses this problem by using unlabeled data together with labeled data in the training process. Co-Training is a popular semi-supervised learning algorithm that has the assumptions that each example is represented by multiple sets of features (views) and these views are sufficient for learning and independent given the class. However, these assumptions are strong and are not satisfied in many real-world domains. In this paper, a single-view variant of Co-Training, called Co-Training by Committee (CoBC) is proposed, in which an ensemble of diverse classifiers is used instead of redundant and independent views. We introduce a new labeling confidence measure for unlabeled examples based on estimating the local accuracy of the committee members on its neighborhood. Then we introduce two new learning algorithms, QBC-then-CoBC and QBC-with-CoBC, which combine the merits of committee-based semi-supervised learning and active learning. The random subspace method is applied on both C4.5 decision trees and 1-nearest neighbor classifiers to construct the diverse ensembles used for semi-supervised learning and active learning. Experiments show that these two combinations can outperform other non committee-based ones.     

Learning model order from labeled and unlabeled data for partially supervised classification, with application to word sense disambiguation

Previous partially supervised classification methods can partition unlabeled data into positive examples and negative examples for a given class by learning from positive labeled examples and unlabeled examples, but they cannot further group the negative examples into meaningful clusters even if there are many different classes in the negative examples. Here we proposed an automatic method to obtain a natural partitioning of mixed data (labeled data + unlabeled data) by maximizing a stability criterion defined on classification results from an extended label propagation algorithm over all the possible values of model order (or the number of classes) in mixed data. Our experimental results on benchmark corpora for word sense disambiguation task indicate that this model order identification algorithm with the extended label propagation algorithm as the base classifier outperforms SVM, a one-class partially supervised classification algorithm, and the model order identification algorithm with semi-supervised k-means clustering as the base classifier when labeled data is incomplete.