DiCE: The Infinitely Differentiable Monte-Carlo Estimator.
Jakob N. FoersterGregory FarquharMaruan Al-ShedivatTim RocktäschelEric P. XingShimon WhitesonPublished in: ICLR (Workshop) (2018)
Keyphrases
- monte carlo
- importance sampling
- variance reduction
- confidence intervals
- markov chain
- monte carlo methods
- least squares
- adaptive sampling
- monte carlo simulation
- finite number
- maximum likelihood
- monte carlo tree search
- markovian decision
- objective function
- particle filter
- policy evaluation
- simulation study
- game tree
- monte carlo method
- maximum a posteriori
- optimal strategy
- estimation error
- game tree search
- global illumination
- matrix inversion
- point processes
- quasi monte carlo
- lower bound