A variance maximization criterion for active learning. (June 2018)

• Record Type: Journal Article
• Authors: Yang, Yazhou; Loog, Marco
• Is Part Of: Pattern recognition. Volume 78(2018:Jun.)
• Page Start: 358
• Page End: 370
• Publication Date: 2018-06
• Subjects: Active learning -- Retraining information matrix -- Variance maximization; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2018.01.017
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A novel active learning approach is proposed, which measures the value of unlabeled data by its predictive variance. To measure the informativeness and representativeness of unlabeled instances, two types of variance are introduced. Excellent performance of the proposed method in comparison with state-of-the-art active learning algorithms is demonstrated. Abstract: Active learning aims to train a classifier as fast as possible with as few labels as possible. The core element in virtually any active learning strategy is the criterion that measures the usefulness of the unlabeled data based on which new points to be labeled are picked. We propose a novel approach which we refer to as maximizing variance for active learning or MVAL for short. MVAL measures the value of unlabeled instances by evaluating the rate of change of output variables caused by changes in the next sample to be queried and its potential labelling. In a sense, this criterion measures how unstable the classifier's output is for the unlabeled data points under perturbations of the training data. MVAL maintains, what we refer to as, retraining information matrices to keep track of these output scores and exploits two kinds of variance to measure the informativeness and representativeness, respectively. By fusing these variances, MVAL is able to select the instances which are both informative and representative. We employ our technique both in combination with logistic regression and support vector machines and demonstrate that MVAL achieves state-of-the-art performance in experiments on a large number of standard benchmark datasets.