Improving Domain Generalization with Interpolation Robustness

Ragja Palakkadavath, Thanh Nguyen-Tang, Sunil Gupta, Svetha Venkatesh

Keywords: domain generalization, limited data, interpolation, invariant representation

TL;DR: We address domain generalization by viewing the underlying distributional shift as interpolation between domains and devise an algorithm to learn a representation that is robustly invariant under such interpolation.

Abstract: We address domain generalization (DG) by viewing the underlying distributional shift as performing interpolation between domains. We devise an algorithm to learn a representation that is robustly invariant under such interpolation and term it as interpolation robustness. We investigate the failure aspect of DG algorithms when availability of training data is scarce. Through extensive experiments, we show that our approach significantly outperforms the recent state-of-the-art algorithm DIRT~\citet{NEURIPS2021_2a271795} and the baseline DeepAll on average across different sizes of data on PACS and VLCS datasets.