Semi-Supervised Novelty Detection

By Gilles Blanchard, Gyemin Lee, Clayton D. Scott. Journal of machine learning research, 2010

313 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

View Paper: https://www.semanticscholar.org/paper/d612d7c21d4130a457968273d79c2c2f6946953d

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 Signal Processing

Citations in the corpus

Deep Anomaly Detection Using Geometric Transformations (2018, background): “This contaminated-sample variant turns out to be easier than the pure version of the problem (in the sense that better performance can be achieved) [2].”
Novelty Detection with GAN (2018, background): “…distribution p g ( x ) that is different from p data ( x ) , can be represented as a unique mixture of p data ( x ) and some other distribution, which by itself cannot be represented as a (nontrivial) mixture of p data ( x ) with another distribution (see Proposition 5 by Blanchard et al. (2010)).”
Learning with Confident Examples: Rank Pruning for Robust Classification with Noisy Labels (2017, background): “Traditional learning performance degrades monotonically with label noise (Aha et al., 1991; Nettleton et al., 2010), necessitating semi-supervised approaches (Blanchard et al., 2010).”
Learning from positive and unlabeled data: a survey (2018, background): “Under this assumption, the optimal classifier can be found by looking for the classifier that classifies all labeled examples as positive and as few as possible examples as negative [63,5].”
Are Anchor Points Really Indispensable in Label-Noise Learning? (2019, background): “[5] Gilles Blanchard, Gyemin Lee, and Clayton Scott.”
Ieee Transactions on Pattern Analysis and Machine Intelligence Classification with Noisy Labels by Importance Reweighting (unknown): “Inspired by the theory of mixture proposition estimation [34], they provided estimators for the inversed noise rates π +1 = P (Y = −1|ˆY = +1) and π −1 = P (Y = +1|ˆY = −1).”
Robust Anomaly Detection and Backdoor Attack Detection Via Differential Privacy (2019, background): “Previous work mostly achieves outlier detection using unsupervised learning methods ([25, 26, 27]), while novelty detection typically assumes a normal dataset is available for training, and is realized by semi-supervised learning ([28, 29]).”
Deep Learning for Anomaly Detection: A Survey (2019, background): “The test instance that does not belong to the majority class is ﬂagged as being anomalous (Perera and Patel [2018], Blanchard et al. [2010]).”
Classification Accuracy as a Proxy for Two Sample Testing (2016, methodology): “A different classifier-based approach to the two-sample problem was proposed by Blanchard et al. (2010). Although their test is built upon classification algorithms, it estimates the a priori probability of a contamination model, instead of classification accuracy.”
Deep Semi-Supervised Anomaly Detection (2019, methodology): “Most existing “semi-supervised” AD methods, both shallow (Muñoz-Marí et al., 2010; Blanchard et al., 2010; Chandola et al., 2009) and deep (Song et al.”
Learning to Locate Relative Outliers (2011, background): “Considering this, several works (Gao et al., 2006; Blanchard et al., 2010; Wu and Ye, 2009) proposed semi-supervised and supervised algorithms for outlier detection.”
Positive-Unlabeled Learning with Non-Negative Risk Estimator (2017, methodology): “…binary classiﬁcation with application to retrieval and nov-elty or outlier detection (Elkan & Noto, 2008; Ward et al., 2009; Scott & Blanchard, 2009; Blanchard et al., 2010), while it also has applications in matrix completion (Hsieh et al., 2015) and sequential data (Li et al., 2009; Nguyen et…”
A Machine Learning Approach to Android Malware Detection (2012, background): “For instance, in [2], the authors reduce the problem of novelty detection in a semi-supervised setting to the problem of Neyman-Pearson Classiﬁcation [29].”
Toward Supervised Anomaly Detection (2014, background): “Blanchard, Lee, and Scott (2010) propose a method which has the appealing option of specifying an upper threshold on the false-positives rate.”
Robust binary hypothesis testing under contaminated likelihoods (2014, background): “2] (originally [8, Prop.”
Convex Formulation for Learning from Positive and Unlabeled Data (2015, methodology): “In practice, it may be estimated with methods such as (Blanchard et al., 2010; du Plessis & Sugiyama, 2014).”
Analysis of Learning from Positive and Unlabeled Data (2014, methodology): “In practice, the unknown class prior π may be estimated by the methods proposed in [10, 1, 11].”