Reparameterization Gradients through Acceptance-Rejection Sampling Algorithms.
Christian A. NaessethFrancisco J. R. RuizScott W. LindermanDavid M. BleiPublished in: AISTATS (2017)
Keyphrases
- computational complexity
- computationally expensive
- computational cost
- learning algorithm
- computational efficiency
- data mining algorithms
- theoretical analysis
- benchmark datasets
- significant improvement
- upper bound
- markov chain
- neural network
- orders of magnitude
- data structure
- monte carlo
- objective function
- image sequences
- times faster
- recently developed
- convergence rate
- gibbs sampler