Decontamination of Training Samples for Supervised Pattern Recognition Methods

By Ricardo Barandela, Eduardo Gasca. 2000

96 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistics and Probability

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 Statistics and Probability

Citations in the corpus

Learning kernel logistic regression in the presence of class label noise (2014, background): “The seemingly straightforward approach is by means of data preprocessing where any suspect samples are removed or relabelled [7,1,29,37,31,18].”
Learning from Noisy Labels with Deep Neural Networks (2014, background): “A simple approach to handle noisy labels is a data preprocessing stage, where labels suspected to be incorrect are removed or corrected [1, 3].”
Training Convolutional Networks with Noisy Labels (2014, methodology): “A simple approach to handle noisy labels is a data preprocessing stage, where labels suspected to be incorrect are removed or corrected (Barandela & Gasca, 2000; Brodley & Friedl, 1999).”
Learning from massive noisy labeled data for image classification (2015, background): “Other data cleansing algorithms are proposed [2, 5, 17], but these approaches are difficult in distinguishing informative hard examples from harmful mislabeled ones.”
Editing Training Data for kNN Classifiers with Neural Network Ensemble (2004, methodology): “In this paper, the Depuration algorithm [2] is examined and two schemes adopted by it are separated so that two new editing approaches are derived.”
Label Noise-Tolerant Hidden Markov Models for Segmentation: Application to ECGs (2011, methodology): “Furthermore, [8] introduces an algorithm to iteratively modify the examples whose class label disagrees with the class labels of most of their neighbours.”
Decoupling "when to update" from "how to update" (2017, methodology): “From the family of preprocessing methods, we mention [4, 10], that try to eliminate instances that are suspected to be mislabeled.”
Analysis of new techniques to obtain quality training sets (2003, methodology): “The first prototype selection technique (Barandela and Gasca, 2000) consists of removing some ‘‘suspicious’’ samples from the TS and also changing the class labels of some other instances.”
Label-Noise Robust Logistic Regression and Its Applications (2012, methodology): “Depuration (denoted as 'Dep' in the figure) tends to perform well in a very high level of noise (i.e. 50% upwards) while at the lower range, its performance is slightly worse.”
Classification in the Presence of Label Noise: A Survey (2014, background): “Learning situations where label noise occurs can be called imperfectly supervised, i.e. pattern recognition applications where the assumption of label correctness does not hold for all the elements of the training sample [29].”