Label-Noise Robust Logistic Regression and Its Applications

By Jakramate Bootkrajang, Ata Kabán. 2012

131 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Molecular Biology

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 Computer Vision and Pattern Recognition

In Molecular Biology

Citations in the corpus

Learning kernel logistic regression in the presence of class label noise (2014, methodology): “Our initial work [4] suggested a more efficient gradient-based algorithm”
Data Programming: Creating Large Training Sets, Quickly (2016, background): “It has also been studied speciﬁcally in the context of label-noise robust logistic regression [5].”
Osprey: Weak Supervision of Imbalanced Extraction Problems without Code (2019, background): “Prior work in weak supervision has focused on the setting of independent crowd workers [7, 9], custom-tailored and hand-tuned distant supervision strategies in the natural language processing domain [14, 21], knowledge-bases[12, 14], handling generic label noise or mis-specification [3, 15].”
Decoupling "when to update" from "how to update" (2017, background): “The effects of noisy labels was vastly studied in many different learning algorithms (see for example the survey in [13]), and various solutions to this problem have been proposed, some of them with theoretically provable bounds, including methods like statistical queries, boosting, bagging and more”
Robust Active Label Correction (2018, methodology): “(3) takes the form as derived by Bootkrajang and Kabán (2012): `ML(σ(x), ỹ) = − log ( (1− ρỹ)σ(ỹx) + ρ−ỹ ( 1− σ(ỹx) )) = `(σ(x), ỹ)− log ( 1 + ρ−ỹ exp(−ỹw>x)− ρỹ ) (7) Unfortunately, minimizing `ML is not a convex problem.”
Loss factorization, weakly supervised learning and label noise robustness (2016, background): “While being out of scope, practical noise estimators are studied [Bootkrajang and Kab´an, 2012, Liu and Tao, 2014, Menon et al., 2015, Scott, 2015].”
Cyberspace Security Using Adversarial Learning and Conformal Prediction (2015, background): “Towards that end, labeling errors characteristic of lacking proper annotation can be detected and redressed in many ways including constrained regularization driven optimization and label flipping [6] with label flipping best done in a principled way characteristic of importance sampling [7].”
Synthetic vs Real: Deep Learning on Controlled Noise (2019, methodology): “We collect noisy labels using image search via Google Image Search, a commonly used way to get noisy labels in prior works (Bootkrajang & Kab´an, 2012; Li et al., 2017a; Krause et al., 2016; Chen & Gupta, 2015; Wang et al., 2014).”
Training deep neural-networks based on unreliable labels (2016, unknown): “[6][7].”
Robust Loss Functions for Learning Multi-class Classifiers (2018, methodology): “Similar methods have been proposed to make logistic regression robust to label noise [11].”
Training deep neural-networks using a noise adaptation layer (2016, methodology): “Variants that are noise robust have been proposed for the most common classifiers such as logisticregression and SVM (Frénay & Verleysen (2014), Jakramate & Kabán (2012), Beigman & Klebanov (2009)).”
Robust Loss Functions under Label Noise for Deep Neural Networks (2017, methodology): “Similar methods have been proposed to make logistic regression robust to label noise [5].”
Snorkel DryBell: A Case Study in Deploying Weak Supervision at Industrial Scale (2018, background): “Prior work in weak supervision has focused on the setting of independent crowd workers [9, 8], custom-tailored and hand-tuned distant supervision strategies in the natural language processing domain [16, 30], and handling generic label noise or misspecification [17, 5], among”