Stochastic Beams and Where To Find Them: The Gumbel-Top-k Trick for Sampling Sequences Without Replacement.
Wouter KoolHerke van HoofMax WellingPublished in: ICML (2019)
Keyphrases
- monte carlo
- tree grammars
- query processing
- hidden markov models
- metropolis hastings
- sampling algorithm
- jump diffusion process
- point processes
- user defined
- random sampling
- stochastic processes
- sequential patterns
- data structure
- parameter space
- information retrieval
- stochastic optimization
- highly relevant
- long sequences
- sampling strategies
- database systems