On the Stratification of Multi-label Data
By Konstantinos Sechidis, Grigorios Tsoumakas, Ioannis P. Vlahavas. 2011
553 citations recorded by the source corpus.
Subfields: Artificial Intelligence, Information Systems, Molecular Biology
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
- Francisco Herrera
- Xavier Serra
- M. J. Jesús
- Eduardo Fonseca
- Min-Ling Zhang
- Jordi Pons
- Francisco Charte
- Xavier Favory
In Information Systems
- Min-Ling Zhang
- Zhi-Hua Zhou
- M. J. Jesús
- Francisco Charte
- Francisco Herrera
- Antonio Jesús Rivera
- Xavier Serra
- Samuel Lipping
In Molecular Biology
- Francisco Herrera
- Min-Ling Zhang
- Zhi-Hua Zhou
- Nils Strodthoff
- Samuel Lipping
- M. J. Jesús
- Alex Alves Freitas
- Tuomas Virtanen
Citations in the corpus
- FSD50K: An Open Dataset of Human-Labeled Sound Events (2020, methodology): “2) val is is computed via iterative stratification [72].”
- Thou shalt not hate: Countering Online Hate Speech (2018, methodology): “For the multi-label classification, we perform multi-label stratified 12 10-fold cross validation (Sechidis, Tsoumakas, and Vlahavas 2011) on the dataset.”
- A Review on Multi-Label Learning Algorithms (2014, background): “When the number of labels is huge, most existing multi-label learning algorithms would suffer from problems of computational inefficiency and class-imbalance [130], [153], especially for the second-order and high-order approaches.”
- Clotho: an Audio Captioning Dataset (2019, methodology): “We use multi-label stratification 4 [10], having as labels for each x o the corresponding words S o , and split D 2000 times in sets of splits of 60%-40%, where 60% corresponds to the development split.”
- Addressing imbalance in multilabel classification: Measures and random resampling algorithms (2015, methodology): “Although BR transformation is maybe more popular when it comes to design new MLC algorithms, LP has been also used to complete other kinds of tasks such as the stratified partitioning of MLDs [36].”
- PTB-XL, a large publicly available electrocardiography dataset (2020, methodology): “In detail, the fold assignment proceeds as follows: The proposed procedure extends existing stratified sampling methods from the literature [26] by accounting for sampling based on patients and by optionally incorporating quality constraints for certain folds.”
- Automated Selection and Configuration of Multi-Label Classification Algorithms with Grammar-Based Genetic Programming (2018, methodology): “We perform the experiments using a stratified five-fold cross-validation [14] with six repetitions varying Auto-MEKAGGP’s random seed, resulting in 30 runs per dataset for each method.”