Domain-Generalisation-in-Invariance

This is the GitHub repository for the paper submitted at ICLR 2023 workshop PML4DC 2023. The link to the paper can be found here: Domain Generalization In Robust Invariant Representation [slides]

The description of different files in the repository can be found below:

• experiments.ipynb - The experiments.ipynb file contains the experiments along with the visualizations of the learned latent manifold for transformations like rotation and translation.
• ood.ipynb - The results can be generated by running the file ood.ipynb which contains the comparisons between the MNIST, FashionMNIST and LFWPeople datasets.
• sample_complexity.ipynb - The sample complexity analysis between Vanilla VAE and Rotation-invariant VAE (rVAE). Here, rVAE refers to the RotInvVAE model discussed in the paper

The skeleton code for the experiments has been borrowed from Explicitly disentangling image content from translation and rotation with spatial-VAE

Citation

@misc{gupta2023domain,
      title={Domain Generalization In Robust Invariant Representation}, 
      author={Gauri Gupta and Ritvik Kapila and Keshav Gupta and Ramesh Raskar},
      year={2023},
      eprint={2304.03431},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}