Identifying and Eliminating Mislabeled Training Instances

By Carla E. Brodley, Mark A. Friedl. 1996

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

Prediction-Based Decisions and Fairness: A Catalogue of Choices, Assumptions, and Definitions (2018, background): “Measurement error in Y is sometimes called noisy labels [127], mislabeled data [17, 16], or unreliable oracles [154].”
Global-Local GCN: Large-Scale Label Noise Cleansing for Face Recognition (2020, background): “Label noise cleansing is a long-term issue in building up a dataset [9, 10, 24, 25, 30, 32].”
Identifying Mislabeled Training Data (1999, methodology): “Results from this work can be found in (Brodley & Friedl, 1996a, 1996b).”
Classification in the presence of class noise using a probabilistic Kernel Fisher method (2007, unknown): “Brodley and Freidl [9] cast this problem into a filtering framework and employed an ensemble of classifiers that served as both filter and final classifier.”
Classification in the Presence of Label Noise: A Survey (2014, methodology): “To solve this problem, ensembles of classiﬁers are used in [46], [138], and [180] to identify mislabeled instances, what is inspired by outlier removal in regression [181].”
Deep learning with noisy labels: exploring techniques and remedies in medical image analysis (2019, methodology): “Voting among an ensemble of classifiers has been shown to be an effective method for this purpose [47], [48].”
Class Noise vs. Attribute Noise: A Quantitative Study (2003, methodology): “A combination of the saturation ﬁlter (Gamberger et al. 2000) and the ﬁltering operation (Brodley and Friedl 1996) was reported in Gamberger et al. (1999), and a Classiﬁcation Filter (CF) scheme was suggested for noise identi-ﬁcation.”
Fast Outlier Detection in High Dimensional Spaces (2002, background): “For example, in classification algorithms mislabelled instances are considered outliers and thus they are removed from the training set to improve the accuracy of the resulting classifier [6].”
Decontamination of Training Samples for Supervised Pattern Recognition Methods (2000, methodology): “Brodley and Friedl [ 9 ] employ a combination of classifiers to filter the training patterns looking for identification and elimination of wrong labeled training cases prior to applying the chosen learning algorithm.”
Map building with mobile robots in dynamic environments (2003, background): “The identification of outliers is an important subtask in various application areas such as data mining [12, 3, 16], correspondence establishment [6, 2], clustering [8], or statistics [1].”
Robust Support Vector Machine Training via Convex Outlier Ablation (2006, background): “Most work focuses on the unsupervised case where there is no designated class variable, but we focus on the supervised case here.”
White Rose Consortium ePrints Repository (unknown): “Brodley and Friedl (Brodley and Friedl, 1996) also use an ensemble approach for identifying misclassiﬁed instances in a training set of pixels from satellite images of land where each pixel needs to be classiﬁed as for example, grassland, woodland etc.”
Identifying and Handling Mislabelled Instances (2004, methodology): “As Brodley and Friedl (1996, 1999) point out, in order to improve classification accuracy, one needs to proceed in two stages: a filtering stage to identify and handle mislabelled examples followed by a prediction stage on the reduced training set.”
Outlier mining in large high-dimensional data sets (2005, unknown): “✦”
Complete Random Forest Based Class Noise Filtering Learning for Improving the Generalizability of Classifiers (2019, methodology): “In order to further reduce the number of samples that are wrongly detected by those methods, a set of classifiers was used in [29], [30] to identify instances of the error labels and generate final class noise by majority voting and consensus voting.”