Resampling Gradients Vanish in Differentiable Sequential Monte Carlo Samplers.
Johannes ZennRobert BamlerPublished in: Tiny Papers @ ICLR (2023)
Keyphrases
- sequential monte carlo
- importance sampling
- markov chain monte carlo
- particle filter
- bayesian inference
- posterior distribution
- visual tracking
- random sampling
- objective function
- monte carlo
- particle filtering
- markov chain
- generative model
- kalman filter
- maximum a posteriori
- object tracking
- appearance model
- hyperparameters
- data association
- em algorithm
- multiple targets
- higher order