An Empirical Study of Invariant Risk Minimization

Code accompanying our paper, An Empirical Study of Invariant Risk Minimization.

This code extends facebookresearch/InvariantRiskMinimization, the official code for the original IRM paper.

Extended ColoredMNIST

See colored_mnist/experiments.sh for scripts that reproduce our results.

PunctuatedSST-2

See punctuated_sst2/experiments.sh for scripts that reproduce our results.

References

If you use this code for research, please cite our paper as follows:

@article{choe2020empirical,
  title={An Empirical Study of Invariant Risk Minimization},
  author={Choe, Yo Joong and Ham, Jiyeon and Park, Kyubyong},
  journal={ICML 2020 Workshop on Uncertainty and Robustness in Deep Learning},
  year={2020}
}