PAC-bayesian analysis of distribution dependent priors: Tighter risk bounds and stability analysis.
Luca OnetoDavide AnguitaSandro RidellaPublished in: Pattern Recognit. Lett. (2016)
Keyphrases
- stability analysis
- bayesian analysis
- risk bounds
- upper bound
- nonlinear systems
- vc dimension
- computationally feasible
- data dependent
- empirical risk minimization
- learning algorithm
- sample complexity
- pac learning
- lower bound
- worst case
- random fields
- neural network
- theoretical framework
- probability distribution
- parameter free
- heavy tailed
- decision making