Removing Undesirable Feature Contributions Using Out-of-Distribution Data

Authors: Saehyung Lee, Changhwa Park, Hyungyu Lee, Jihun Yi, Jonghyun Lee, Sungroh Yoon

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

Reproducibility Variable Result LLM Response
Research Type Experimental We show how to improve generalization theoretically using OOD data in each learning scenario and complement our theoretical analysis with experiments on CIFAR-10, CIFAR-100, and a subset of Image Net.
Researcher Affiliation Academia Saehyung Lee, Changhwa Park, Hyungyu Lee, Jihun Yi, Jonghyun Lee, Sungroh Yoon Electrical and Computer Engineering, AIIS, ASRI, INMC, and Institute of Engineering Research Seoul National University Seoul 08826, South Korea
Pseudocode Yes Algorithm 1 Out-of-distribution augmented Training (OAT)
Open Source Code Yes Our codes are available at https://github.com/Saehyung-Lee/OAT.
Open Datasets Yes experiments on CIFAR-10 (Krizhevsky et al., 2009), CIFAR-100 (Krizhevsky et al., 2009), and a subset of Image Net (Deng et al., 2009).
Dataset Splits No The paper mentions training and test set sizes (e.g., "Img Net10 (train set size = 9894 and test set size = 3500)"), but does not explicitly provide details for a separate validation split, nor does it refer to standard splits that include a validation set.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or cloud instance types) used for running the experiments.
Software Dependencies No The paper mentions general software like "neural networks" or "CNNs" but does not specify any particular software libraries, frameworks, or their version numbers that are required for reproducibility.
Experiment Setup Yes For all the experiments except TRADES and OATTRADES, the initial learning rate is set to 0.1. The learning rate is multiplied by 0.1 at 50% and 75% of the total training steps, and the weight decay factor is set to 2e 4. We use the same adversarial perturbation budget ϵ = 8, as in Madry et al. (2017).