Full Professor, École Polytechnique
2 papers at NeurIPS 2025
When multiple conformal predictors are available, we allow data-dependent selection without the loss of coverage.