@misc{moreno-torres_tackling_nodate,
title = {Tackling dataset shift in classification: Benchmarks and methods},
author = {José G Moreno-Torres and Troy Raeder and Rocío Alaiz-Rodríguez and Nitesh V Chawla and Francisco Herrera},
url = {https://scholar.google.es/citations?view_op=view_citation&hl=en&user=2gj1UNYAAAAJ&cstart=20&pagesize=80&sortby=title&citation_for_view=2gj1UNYAAAAJ:0EnyYjriUFMC},
abstract = {This paper addresses the issue of dataset shift, which occurs when the data used to train a classifier differs from the data distribution it encounters during deployment. This phenomenon can lead to poor performance of the classifier, similar to the impact of noisy data. The paper introduces a new benchmark set of datasets to facilitate fair comparisons of algorithms designed to handle dataset shift. The study also includes a comprehensive analysis of key algorithms in the field, evaluating their effectiveness across a range of datasets and shifts.},
keywords = {Algorithm Comparison, Benchmark Dataset, Classification, dataset shift, Transfer Learning},
pubstate = {published},
tppubtype = {misc}
}
This paper addresses the issue of dataset shift, which occurs when the data used to train a classifier differs from the data distribution it encounters during deployment. This phenomenon can lead to poor performance of the classifier, similar to the impact of noisy data. The paper introduces a new benchmark set of datasets to facilitate fair comparisons of algorithms designed to handle dataset shift. The study also includes a comprehensive analysis of key algorithms in the field, evaluating their effectiveness across a range of datasets and shifts.