Oops I Took A Gradient: Scalable Sampling for Discrete Distributions.
Will GrathwohlKevin SwerskyMilad HashemiDavid DuvenaudChris J. MaddisonPublished in: ICML (2021)
Keyphrases
- efficient algorithms to compute
- random sampling
- power law
- sampling algorithm
- sampled data
- monte carlo
- random samples
- gaussian distribution
- web scale
- sampling methods
- discrete version
- highly skewed
- sampling strategies
- band limited
- gradient method
- gradient information
- zero crossing
- joint distribution
- sample size
- lightweight
- edge detection
- probability distribution
- scalable video coding
- translation invariant
- data sets
- highly scalable
- heavy tailed
- discrete variables
- discrete geometry
- parameter space
- multivariate gaussian