Committee-Based Sampling For Training Probabilistic Classifiers

By Ido Dagan, Shlomo Engelson Argamon. 1995

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

Replay in Deep Learning: Current Approaches and Missing Biological Elements (2021, neutral): “…Wang, Zhang, Li, Zhang, & Lin, 2016; Settles, 2009; Yoo & So Kweon, 2019; Wei, Iyer, & Bilmes, 2015), with the most common selection methods using uncertainty sampling (Lewis & Gale, 1994; Culotta & McCallum, 2005; Scheffer, Decomain, & Wrobel, 2001; Dagan & Engelson, 1995; Dasgupta & Hsu, 2008).”
Scaling to Very Very Large Corpora for Natural Language Disambiguation (2001, background): “For example, Dagan and Engelson (1995) describe a committee-based sampling technique where a part of speech tagger is trained using an annotated seed corpus.”
Self-Supervised Exploration via Disagreement (2019, background): “Common selection criteria include entropy (Dagan & Engelson, 1995), uncertainty sampling (Lewis & Gale, 1994) and expected informative-ness (Houlsby et al., 2011).”
Active and Semisupervised Learning for the Classification of Remote Sensing Images (2014, background): “Another strategy is query by committee (QBC) [36], [37].”
SeqMix: Augmenting Active Sequence Labeling via Sequence Mixup (2020, background): “, 2006) and committee-based approaches (Dagan and Engelson, 1995) have been proposed for improving the sample efficiency.”
Democratic co-learning (2004, unknown): “Instead of basing priorities on the number of disagreements, we consider a variant [15] of QBC where the priority of example Ü is computed using the entropy of the classifications voted by each member where Ò Ø Ö Ó Ô Ý ´Üµ µ È Ö ½ Î ´µ ÐÓÓ ¾ Î ´µ for the number of committee members, Ö the total…”
Diverse ensembles for active learning (2004, methodology): “In their QBC approach, Dagan and Engelson (1995) measure the utility of examples by vote entropy, which is the entropy of the class distribution, based on the majority votes of each committee member.”
Selective Sampling Using the Query by Committee Algorithm (1997, background): “Though theoretical results for such models are lacking, there is empirical evidence that extensions of the QBC algorithm can be used to learn noisy and probabilistic models, such as hidden Markov models (Dagan & Engelson,1995).”
Active Learning Literature Survey (2009, methodology): “The ﬁrst is vote entropy (Dagan and Engelson, 1995): where y i again ranges over all possible labelings, and V ( y i ) is the number of “votes” that a label receives from among the committee members’ predictions.”
A machine learning approach to sentiment analysis in multilingual Web texts (2009, methodology): “Good overviews of AL are found in Dagan and Engelson (1995) and Baram et al. (2003).”
Selective Sampling In Natural Language Learning (1995, methodology): “Some details of our implementation are given below; the system is more fully described in (Dagan and Engelson, 1995 ).”
Employing EM and Pool-Based Active Learning for Text Classification (1998, methodology): “Some previous applications of QBC [Dagan & Engelson 1995; Liere & Tadepalli 1997] use a simulated stream of unlabeled documents.”
Text Classification from Labeled and Unlabeled Documents using EM (2000, background): “Three such examples applied to text are “Query By Committee” ( Dagan & Engelson, 1995; Liere & Tadepalli, 1997), relevance sampling and uncertainty sampling (Lewis & Gale, 1994; Lewis, 1995).”
Active Imitation Learning via Reduction to I.I.D. Active Learning (2012, methodology): “We use the entropy of the vote distribution (Dagan and Engelson 1995) as a typical measure of committee disagreement.”
An Analysis of Active Learning Strategies for Sequence Labeling Tasks (2008, methodology): “, 2006), or query-by-committee, where a “committee” of models selects the instance about which its members most disagree (Dagan and Engelson, 1995).”
Cost-Effective Incremental Deep Model: Matching Model Capacity With the Least Sampling (2023, unknown): “reducing the uncertainty [21], [22]; and 2) Representativeness, which measures the representation of an instance to the overall input patterns [23].”