Learning from labeled features using generalized expectation criteria

By Gregory Druck, Gideon S. Mann, Andrew McCallum. 2008

291 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Representation & Self-Supervised Learning

How researchers see this paper

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

In Artificial Intelligence

In Representation & Self-Supervised Learning

Citations in the corpus

Concrete Problems in AI Safety (2016, background): “For example, a quadcopter sufficiently far from the ground might be able to explore safely, since even if something goes wrong there will be ample time for a human or another policy to rescue it.”
Model-Agnostic Interpretability of Machine Learning (2016, background): “Feature labeling (Druck et al., 2008) or annotator rationales (Zaidan & Eisner, 2008) are other forms of feedback that should be supported for explanations.”
Modeling Annotators: A Generative Approach to Learning from Annotator Rationales (2008, background): “Druck et al. (2008) show annotators some features f from a ﬁxed feature set, and ask them to choose a class label y such that p ( y | f ) is as high as possible.”
Generalized Expectation Criteria for Semi-Supervised Learning of Conditional Random Fields (2008, methodology): “When these distributions are estimated by means of annotated feature occurrences in context, there is as much as a ten-fold reduction in the annotation time that is required in order to achieve the same level of accuracy over traditional instance-labeling.”
Consistent Multiclass Algorithms for Complex Performance Measures (2015, methodology): “Performance of Bisection on IR Data Sets Next, we evaluate the bisection based algorithm on three information retrieval (IR) data sets, where the micro F1measure is of interest: a version of the CoRA data set containing research papers categorized into 7 classes, the 20 Newsgroups data set containin”
Trains of thought: generating information maps (2012, methodology): “[5] proposed a discriminative semi-supervised learning method that incorporates into training affinities between features and classes.”
Active Learning Literature Survey (2009, methodology): “In recent years, several new methods have been developed for incorporating feature-based domain knowledge into supervised and semi-supervised learning (e.g., Haghighi and Klein, 2006; Druck et al., 2008).”
The Constrained Weight Space SVM: Learning with Ranked Features (2011, methodology): “While methods for learning with labeled features have been recently proposed elsewhere (e.g., (Zaidan et al., 2007; Druck et al., 2008), which we discuss at length in Section 5), our method is unique in two ways.”
Interactive Annotation Learning with Indirect Feature Voting (2009, background): “Supervision for simple features has been explored in the literature (Raghavan et al., 2006; Druck et al., 2008; Haghighi and Klein, 2006).”
Interactive Feature Space Construction using Semantic Information (2009, methodology): “These methods allow the annotator to directly specify information about the feature space in addition to providing labels, which is then incorporated into the learning algorithm (Huang and Mitchell, 2006; Raghavan and Allan, 2007; Zaidan et al., 2007; Druck et al., 2008; Zaidan and Eisner, 2008).”
Sentiment analysis of blogs by combining lexical knowledge with text classification (2009, background): “[7] incorporate prior knowledge through labeled features, which are used to directly constrain the model’s predictions on unlabeled instances.”
Zero-shot Learning of Classifiers from Natural Language Quantification (2018, methodology): “Each dot represents a dataset generated from a known distribution. that uses the Feature Labeling through Generalized Expectation criterion, following the approach in Druck et al. (2008); Mann and McCallum (2010).”
Active Learning by Labeling Features (2009, methodology): “A similar heuristic was used by Druck et al. (2008).”
Generalized Expectation Criteria for Semi-Supervised Learning with Weakly Labeled Data (2010, background): “Second, Druck et al. (2008) reports results of experiments using GE criteria where the input distributions were only allowed to affect the features th y were conditioned on (corresponding to only being able to adjust the label proportions).”
Explanatory Interactive Machine Learning (2019, unknown): “Morerecently,researchersexploredfeaturesupervision[1,10, 11,23,24,30]andrationales[32,40,41],whichleveragebothlabel-andfeature-level(orsentence-level,forrationales)supervisionwith theaimofimprovinglearningefficiency.”
Distribution Matching for Transduction (2009, background): “Corresponding distributional assumptions are made for classification by [5], for regression by [10], and in the context of sufficient statistics on the marginal polytope by [3, 6].”