PF-GNN: Differentiable particle filtering based approximation of universal graph representations.
Mohammed Haroon DuptyYanfei DongWee Sun LeePublished in: ICLR (2022)
Keyphrases
- particle filtering
- particle filter
- graph representations
- bayesian filtering
- visual tracking
- object tracking
- appearance model
- graph structure
- kalman filter
- mean shift
- proposal distribution
- graph theory
- graph representation
- state space
- human motion
- graph kernels
- objective function
- pairwise
- bayesian inference
- learning algorithm
- structured data
- face recognition
- feature space
- density estimation
- machine learning
- graphical models