1 paper
Nick Masi, Randall Balestriero
The dominant paradigm in machine learning is to assess model performance based on average loss across all samples in some test set. This amounts to averaging performance geospatial…