PVEs: Position-Velocity Encoders for Unsupervised Learning of Structured State Representations.
Rico JonschkowskiRoland HafnerJonathan ScholzMartin A. RiedmillerPublished in: CoRR (2017)
Keyphrases
- unsupervised learning
- supervised learning
- constant velocity
- semi supervised
- position and orientation
- deep learning
- object recognition
- structured data
- dimensionality reduction
- angular velocity
- video compression
- expectation maximization
- optical flow
- machine learning
- active learning
- relative position
- bayesian networks
- case study
- feature selection