Noise Models for Classification Datasets
Asymmetric default label noise
Asymmetric interval-based attribute noise
Asymmetric sparse label noise
Asymmetric uniform attribute noise
Asymmetric uniform label noise
Attribute-mean uniform label noise
Distance to SVM decision boundary
Mislabeling based on k-nearest neighbors
Boundary/dependent Gaussian attribute noise
Clustering-based voting label noise
Exponential borderline label noise
Exponential/smudge completely-uniform label noise
Find the differences between two datasets
Fraud bidirectional label noise
Gamma borderline label noise
Gaussian borderline label noise
Gaussian-mixture borderline label noise
Gaussian-level uniform label noise
Hubness-proportional uniform label noise
Importance interval-based attribute noise
IR-stable bidirectional label noise
Laplace borderline label noise
Large-margin uniform label noise
Majority-class unidirectional label noise
Minority-driven bidirectional label noise
Minority-proportional uniform label noise
Misclassification prediction label noise
Multiple-class unidirectional label noise
Neighborwise borderline label noise
Non-linearwise borderline label noise
Type of noise introduced by a noise model
One-dimensional uniform label noise
Open-set ID/nearest-neighbor label noise
Open-set ID/uniform label noise
Pairwise bidirectional label noise
Plot function for class ndmodel
PMD-based confidence label noise
Print function for class ndmodel
Print function for class sum.ndmodel
Quadrant-based uniform label noise
Random numbers considering reference values
Safe sample function
Sample considering reference values
Score-based confidence label noise
Sigmoid-bounded uniform label noise
Small-margin borderline label noise
Smudge-based completely-uniform label noise
Summary function for class ndmodel
Symmetric adjacent label noise
Symmetric center-based label noise
Symmetric confusion label noise
Symmetric completely-uniform attribute noise
Symmetric completely-uniform combined noise
Symmetric completely-uniform label noise
Symmetric double-default label noise
Symmetric default label noise
Symmetric diametrical label noise
Symmetric double-random label noise
Symmetric end-directed attribute noise
Symmetric exchange label noise
Symmetric Gaussian attribute noise
Symmetric hierarchical label noise
Symmetric hierarchical/next-class label noise
Symmetric interval-based attribute noise
Symmetric natural-distribution label noise
Symmetric nearest-neighbor label noise
Symmetric next-class label noise
Symmetric non-uniform label noise
Symmetric optimistic label noise
Symmetric pessimistic label noise
Symmetric scaled-Gaussian attribute noise
Symmetric uniform attribute noise
Symmetric uniform label noise
Symmetric unit-simplex label noise
Symmetric/dependent Gaussian attribute noise
Symmetric/dependent Gaussian-image attribute noise
Symmetric/dependent random-pixel attribute noise
Symmetric/dependent uniform attribute noise
Uneven-Gaussian borderline label noise
Uneven-Laplace borderline noise
Unconditional fixed-width attribute noise
Unconditional vp-Gaussian attribute noise
Unconditional/symmetric Gaussian/uniform combined noise
Implementation of models for the controlled introduction of errors in classification datasets. This package contains the noise models described in Saez (2022) <doi:10.3390/math10203736> that allow corrupting class labels, attributes and both simultaneously.