A unifying view on dataset shift in classification

By Jose Garcia Moreno-Torres, Troy Raeder, R. Alaíz-Rodríguez, Nitesh V. Chawla, Francisco Herrera. 2012

1,155 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Safety, Robustness & Interpretability, Statistical Learning Theory

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 Safety, Robustness & Interpretability

In Statistical Learning Theory

Citations in the corpus

Advances and Open Problems in Federated Learning (2019, background): “This taxonomy has a close mapping to notions of dataset shift [304, 327], which studies differences between the training distribution and testing distribution; here, we consider differences in the data distribution on each client.”
Characterizing Fairness Over the Set of Good Models Under Selective Labels (2021, background): “, Chapter 12 of Imbens & Rubin (2015) and Heckman (1990))3 and in covariate shift learning (Moreno-Torres et al., 2012).”
Learning to Generate Novel Domains for Domain Generalization (2020, neutral): “Typically, machine-learning models perform poorly when deployed on test data with a different data distribution than the training data, which is known as the domain shift problem [35].”
SMOTE for Learning from Imbalanced Data: Progress and Challenges, Marking the 15-year Anniversary (2018, background): “In particular, we may find in the work of Moreno-Torres et al. (2012b) a novel approach that is not biased to this problem.”
Data augmentation using generative adversarial networks (CycleGAN) to improve generalizability in CT segmentation tasks (2019, neutral): “Volume (2) Volume ExpertSegmentation UNetSegmentation”
Noise as a Resource for Learning in Knowledge Distillation (2020, background): “However, models in the real world often have to deal with some form of domain shift which adversely affects the generalization performance of the models [25, 33, 39, 49].”
A review of microarray datasets and applied feature selection methods (2014, background): “This occurs when the testing (unseen) data experience a phenomenon that leads to a change in the distribution of a single feature, a combination of features, or the class boundaries [83].”
Machine Learning for 5G/B5G Mobile and Wireless Communications: Potential, Limitations, and Future Directions (2019, methodology): “parameters of the model accordingly to the changes and retraining the algorithm with an augmented or modified dataset [42].”
A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, neutral): “There are two major assumptions of domain shift [1], [27]: covariate shift in which the features cause the labels; and label shift in which the labels cause the features.”
Online Machine Learning in Big Data Streams (2018, background): “A few papers [168, 171] give overviews of different types of environmental changes and concept drifts based on speed, recurrence, and severity.”
Learn from Others and Be Yourself in Heterogeneous Federated Learning (2022, neutral): “A myriad of methods [43,46,73,77] incorporate extra proximal terms to handle the data in label distribution skew (prior probability shift) [30], neglecting the fact that there exists domain shift (same label, different features) [60, 64, 66].”
Analyzing concept drift and shift from sample data (2018, background): “…and Kubat 1996; Kifer et al. 2004; Gama et al. 2004; Baena-Garcıa et al. 2006; Nishida and Yamauchi 2007; Dries and Rückert 2009; Gama and Rodrigues 2009; Žliobaite 2010; Hoens et al. 2012; Moreno-Torres et al. 2012; Bifet et al. 2013; Gama et al. 2014; Qahtan et al. 2015; Yu and Abraham 2017).”
TESSERACT: Eliminating Experimental Bias in Malware Classification across Space and Time (2018, background): “[31] systematize different drifts, and mention sample-selection bias; while this resembles spatial bias, [31] does not propose any solution/experiments for its impact on ML performance.”