Selective Sampling Using the Query by Committee Algorithm

By Yoav Freund, H. Sebastian Seung, E. Shamir, Naftali Tishby. 1997

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Reliable Uncertainty Estimates in Deep Neural Networks using Noise Contrastive Priors (2018, background): “Ensemble methods (Seung et al., 1992; McCallumzy and Nigamy, 1998; Freund et al., 1997) generate queries that have the greatest disagreement between a set of classiﬁers.”
Multi-Task Active Learning for Linguistic Annotations (2008, methodology): “committee-based (Freund et al., 1997) selection protocols.”
Hierarchical sampling for active learning (2008, background): “Indeed, there are canonical examples in which active learning provably yields exponentially lower label complexity than supervised learning (Cohn et al., 1994; Freund et al., 1997; Dasgupta, 2005; Balcan et al., 2006; Balcan et al., 2007; Castro & Nowak, 2007; Hanneke, 2007; Dasgupta et al., 2007).”
Combining Committee-Based Semi-Supervised Learning and Active Learning (2010, methodology): “Query by Committee (QBC)[14] is a committee-based active learning algorithm, in which an ensemble of diverse classifiers is constructed.”
Hyperparameter Learning for Graph Based Semi-supervised Learning Algorithms (2006, background): “(Freund et al., 1997) showed theoretically that if there is no error in labelling, QBC can exponentially reduce the number of labelled examples needed for learning.”
Adaptive Active Learning for Image Classification (2013, methodology): “Query-by-committee algorithms train a committee of classifiers and choose the instance on which the committee members most disagree [9, 19].”
Semi-Automated Data Labeling (2020, background): “Popular selection heuristics include uncertainty, expected error reduction Roy and McCallum (2001) and query by committee Freund et al. (1997).”
Active Learning Literature Survey (2009, background): “The main theoretical result to date in the stream-based and pool-based scenarios seems to be an analysis of the query-by-committee (QBC) algorithm by Freund et al. (1997).”
A Survey of Active Learning Algorithms for Supervised Remote Sensing Image Classification (2011, background): “The first family of active learning methods quantifies the uncertainty of a pixel by considering a committee of learners [23], [24].”
Query by Committee Made Real (2005, background): “[6] proved that this is the expected behavior.”
Batch mode active learning and its application to medical image classification (2006, methodology): “To reduce the effort involved in acquiring labeled examples, a number of active learning methods (Fine et al., 2002; Freund et al., 1997; Graepel & Herbrich, 2000; Seung et al., 1992; Campbell et al., 2000; Schohn & Cohn, 2000; Tong & Koller, 2000) Appearing in Proceedings of the 23 rd…”
Interactive deduplication using active learning (2002, methodology): “Active learning [7, 9] methods also rely on a similar insight for selecting instances for labeling from a large pool of unlabeled instances.”
Multilabel Classification using Bayesian Compressed Sensing (2012, background): “Several criteria, such as uncertainty [12], expected informativeness [13, 14], reduction in version space [15], disagreement among a committee of classifiers [16], etc.”
Active Feature-Value Acquisition (2009, background): “Techniques from optimal experimental design (Federov 1972) and from active learning (Cohn et al. 1994, Freund et al. 1997) assume class labels are unknown.”
Semi-Supervised Classification of Network Data Using Very Few Labels (2010, methodology): “It is similar to many recent graph-based SSL methods (e.g., [2], [3]) and is shown to be essentially the same as the Gaussian-field classifier proposed by Zhu et al. [4] and proves to be very effective on some benchmark network datasets.”
Active Semi-Supervision for Pairwise Constrained Clustering (2004, background): “, reduction of the version space size [16], reduction of uncertainty in predicted label [24], maximizing the margin on training data [1], finding high variance data points by density-weighted pool-based sampling [27], etc.”