Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design
Adam Foster * 1 Desi R. Ivanova * 1 Ilyas Malik 2 Tom Rainforth 1
Abstract using a likelihood p(y|θ, ξ) and a prior p(θ), where ξ is
We introduce Deep Adaptive Design (DAD), our controllable design and θ is the set of parameters we
a method for amortizing the cost of adaptive wish to learn about. We then optimize ξ to maximize the ex-
Bayesian experimen ...


雷达卡




京公网安备 11010802022788号







