Embodiment-specific representation of robot grasping using graphical models and latent-space discretization.
Dan SongCarl Henrik EkKai HuebnerDanica KragicPublished in: IROS (2011)
Keyphrases
- graphical models
- latent space
- latent variables
- probabilistic model
- random variables
- vision system
- gaussian process
- low dimensional
- generative model
- lower dimensional
- latent variable models
- parameter space
- feature space
- conditional random fields
- bayesian networks
- probabilistic latent semantic analysis
- matrix factorization
- manifold learning
- dimensionality reduction
- high dimensional data
- distance metric
- gaussian mixture
- hidden variables
- gaussian processes
- high dimensional
- bayesian framework
- human motion
- transfer learning
- cross validation
- image segmentation