Active Learning with Feedback on Features and Instances

By Hema Raghavan, Omid Madani, Rosie Jones. 2006

216 citations recorded by the source corpus.

How researchers see this paper

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

In Statistical Learning Theory

Citations in the corpus

Rewriting a Deep Generative Model (2020, background): “Interactive machine learning systems aim to improve training through human interaction in labeling [15,22,68], or by allowing a user to to aid in the model optimization process via interactive feature selection [19,27,64,50] or model”
Modeling Annotators: A Generative Approach to Learning from Annotator Rationales (2008, methodology): “Thus, Raghavan et al. (2006) propose an active learning method in which, while classifying a training document, the annotator also identifies some features of that document as particularly relevant.”
Where Are My Intelligent Assistant's Mistakes? A Systematic Testing Approach (2011, background): “that allowing users to debug by labeling a word/phrase is up five times more efficient than simply labeling training messages [23].”
Interactive Annotation Learning with Indirect Feature Voting (2009, methodology): “Supervision for simple features has been explored in the literature (Raghavan et al., 2006; Druck et al., 2008; Haghighi and Klein, 2006).”
Closing the Loop: Fast, Interactive Semi-Supervised Annotation With Queries on Features and Instances (2011, methodology): “Feature queries were less costly than instances, which is consistent with findings in previous work (Raghavan et al., 2006; Druck et al., 2009).”
A Review of User Interface Design for Interactive Machine Learning (2018, background): “Allowing users to select features has been found to deliver both efficiency (Raghavan et al. 2006) and quality (Brooks et al. 2015) gains.”
Learning from labeled features using generalized expectation criteria (2008, unknown): “Raghavan, et al. [18] perform a thorough user study in which they conclude that it is ﬁve times faster to label a feature than to label a document.”
You Are the Only Possible Oracle: Effective Test Selection for End Users of Interactive Machine Learning Systems (2014, background): “Other researchers have similarly reported that allowing users to debug by labeling a word/phrase is up to five times more efficient than simply labeling training messages [46].”
Active Learning Literature Survey (2009, background): “Raghavan et al. (2006) have proposed a related idea for traditional classiﬁcation problems called tandem learning , in which the learner is allowed to query for the labels of features as well as entire instances.”
From Theories to Queries (2011, background): “…reason about annotation cost, several authors have found that alternative query types (such as labeling features rather than instances, see Section 2.4) can lead to reduced annotation costs for human oracles (Raghavan et al., 2006; Druck et al., 2009; Vijayanarasimhan and Grauman, 2009a).”