Introduces an information-matching approach based on the Fisher Information Matrix for optimal experimental design and active learning to select informative training data for models with sloppy parameters.
Conic optimization via operator splitting and homoge- neous self-dual embedding
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
An information-matching approach to optimal experimental design and active learning
Introduces an information-matching approach based on the Fisher Information Matrix for optimal experimental design and active learning to select informative training data for models with sloppy parameters.