Revisiting Out-of-Distribution Generalization

Explore out-of-distribution generalization in machine learning through this conference talk from the Computational Genomics Summer Institute (CGSI) 2023. Delve into the challenges of spurious correlations and their impact on model performance across different domains. Examine three related papers that address nuisance-induced spurious correlations, data augmentation techniques for adjusting these correlations, and the relationship between robustness to spurious correlations and semantic out-of-distribution detection. Gain insights into cutting-edge research aimed at improving machine learning models' ability to generalize beyond their training distribution, with potential applications in computational genomics and beyond.

Rajesh Ranganath | Revisiting Out of Distribution Generalization | CGSI 2023