Selecting Data Augmentation for Simulating Interventions

Authors: Maximilian Ilse, Jakub M Tomczak, Patrick Forré

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4. Experiments We evaluate the performance of data augmentation in combination with Empirical Risk Minimization (ERM) (Vapnik, 1992) on four datasets. While the first is a synthetic dataset, the other three are domain generalization benchmark image datasets (rotated MNIST, colored MNIST, and PACS) where the domain d and target y are confounded.
Researcher Affiliation Academia 1Amsterdam Machine Learning Lab, University of Amsterdam 2Computational Intelligence Group, Vrije Universiteit Amsterdam.
Pseudocode No The paper describes the proposed SDA algorithm in a numbered list using prose, but it does not provide structured pseudocode or an algorithm block.
Open Source Code Yes Code to replicate all experiments can be found under https://github.com/AMLab-Amsterdam/ Data Augmentation Interventions.
Open Datasets Yes We evaluate the performance of data augmentation in combination with Empirical Risk Minimization (ERM) (Vapnik, 1992) on four datasets. While the first is a synthetic dataset, the other three are domain generalization benchmark image datasets (rotated MNIST, colored MNIST, and PACS) where the domain d and target y are confounded. The PACS dataset (Li et al., 2017a) was introduced as a strong benchmark dataset for domain generalization methods...
Dataset Splits Yes 1. We divide all samples from the training domains into a training and validation set.
Hardware Specification No The paper does not provide specific hardware details (such as GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper mentions machine learning models and general frameworks but does not provide specific version numbers for any software libraries, dependencies, or programming languages used (e.g., 'PyTorch 1.9').
Experiment Setup Yes The hyperparameter for each augmentation can be found in the Appendix. In addition, we perform an ablation study showing that SDA reliably picks the most suitable hyperparameters, the results can be found in Table 4 in the Appendix.