← Search

Lorenzo Beretta

2 accepted papers

2024

Sketched Lanczos uncertainty score: a low-memory summary of the Fisher information

NeurIPS 2024poster

Current uncertainty quantification is memory and compute expensive, which hinders practical uptake. To counter, we develop Sketched Lanczos Uncertainty (SLU): an architecture-agnostic uncertainty score that can be applied to pre-trained neural networks with minimal overhead. Importantly, the memory…

Cited by 2SourcePDFScholar