Tom Rainforth
3 indexed papers
Publications per year
Top categories
Frequent co-authors
Research Timeline
This paper proposes a new formulation of Bayesian experimental design (BED) based on expected future loss (EFL) on downstream actions, which can be optimized using stochastic gradients and does not require explicit posterior or marginal likelihood estimation.
The paper introduces FairBED, a method to gather fairer data through Bayesian experimental design that minimizes information about sensitive attributes.
This paper introduces POLAR, a unified framework for adaptive data acquisition that leverages pretrained models as belief-state encoders.
Papers
Efficient Adaptive Data Acquisition via Pretrained Belief Representations
Daolang Huang, Zhuoyue Huang, Conor Hassan, Luigi Acerbi +2 more
This paper introduces POLAR, a unified framework for adaptive data acquisition that leverages pretrained models as belief-state encoders.