PAC-Bayesian Margin Bounds for Convolutional Neural Networks - Technical Report.
Konstantinos PitasMike E. DaviesPierre VandergheynstPublished in: CoRR (2018)
Keyphrases
- pac bayesian
- technical report
- convolutional neural networks
- rademacher complexity
- distribution free
- data dependent
- convolutional network
- generalization bounds
- error bounds
- decision boundary
- risk bounds
- support vector
- objective function
- training set
- statistical learning theory
- large deviations
- covariance matrix
- theoretical analysis