The Missing Invariance Principle found -- the Reciprocal Twin of Invariant Risk Minimization

Authors: Dongsung Huh, Avinash Baidya

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We also empirically demonstrate that MRI-v1 strongly out-performs IRM-v1 and consistently achieves near-optimal OOD generalization in image-based nonlinear problems.
Researcher Affiliation Collaboration Dongsung Huh MIT-IBM Watson AI Lab Cambridge, MA 02142 huh@ibm.com Avinash Baidya Department of Physics and Astronomy University of California Davis, CA 95616 aavinash@ucdavis.edu
Pseudocode No The paper does not contain a clearly labeled 'Pseudocode' or 'Algorithm' block.
Open Source Code Yes Code available at https://github.com/IBM/MRI.
Open Datasets Yes Colored MNIST (CMNIST) CMNIST (Arjovsky et al., 2019) is a synthetic dataset derived from MNIST for binary classification.
Dataset Splits No The paper mentions 'two training environments Etr = {e1, e2}' and 'testing environment Etest = {e0}' but does not explicitly provide specific training/validation/test dataset splits (e.g., percentages or sample counts) within these environments in the main text.
Hardware Specification No The paper states that compute resources are detailed in the Supplementary Materials but does not provide specific hardware details (e.g., GPU/CPU models, memory) in the main text.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., 'PyTorch 1.9', 'Python 3.8') in the main text.
Experiment Setup No The paper mentions testing algorithms 'under a wide range of hyperparameters' and refers to 'Supplementary Materials' for 'training details (e.g., data splits, hyperparameters, how they were chosen)', but does not provide concrete hyperparameter values or detailed experimental setup in the main text.