Bandit Theory meets Compressed Sensing for high dimensional Stochastic Linear Bandit.
Alexandra CarpentierRémi MunosPublished in: AISTATS (2012)
Keyphrases
- compressed sensing
- regret bounds
- high dimensional
- image reconstruction
- random projections
- random sampling
- sparse representation
- natural images
- low dimensional
- compressive sampling
- lower bound
- nearest neighbor
- signal processing
- dimensionality reduction
- similarity search
- upper bound
- high dimensionality
- orthogonal matching pursuit
- image classification
- data points
- active learning
- feature space
- pattern recognition
- image processing