PAC-Bayes Generalisation Bounds for Heavy-Tailed Losses through Supermartingales.
Maxime HaddoucheBenjamin GuedjPublished in: Trans. Mach. Learn. Res. (2023)
Keyphrases
- pac bayes
- heavy tailed
- risk bounds
- large deviations
- generalization bounds
- linear classifiers
- generalized gaussian
- data dependent
- learning theory
- vc dimension
- empirical risk minimization
- stochastic processes
- statistical learning theory
- image segmentation
- generalization error
- distance measure
- markov random field
- upper bound
- support vector