PAC-Bayes 的上界可以理解为是经验误差和模型复杂度之间的均衡，经验误差反映了模型拟合数据的程度，而复...
PAC-Bayes 的上界可以理解为是经验误差和模型复杂度之间的均衡，经验误差反映了模型拟合数据的程度，而复...
We prove a general PAC-Bayesian bound, and show how to use it in various hostile settings.doi:10.1007/s10994-017-5690-0Pierre AlquierBenjamin GuedjSpringer USMachine LearningP. Alquier and B. Guedj. Simpler pac-bayesian bounds for hostile data. Machine Learning, 107(5):887-902, 2018....
The PAC-Bayesian boundnaturally handles infinite precision rule parameters, $L_2$ regularization,{\\em provides a bound for dropout training}, and defines a natural notion of asingle distinguished PAC-Bayesian posterior distribution. The third bound is atraining-variance bound --- a kind of bias...
To this end, we propose a method to learn GPs and their sparse approximations by directly optimizing a PAC-Bayesian bound on their generalization performance, instead of maximizing the marginal likelihood. Besides its theoretical appeal, we find in our evaluation that our learning method is robust ...
We provide a PAC-Bayesian bound for the expected loss of convex combinations of classifiers under a wide class of loss functions (which includes the exponential loss and the logistic loss). Our numerical experiments with Adaboost indicat... B Schölkopf,J Platt,T Hofmann - MIT Press 被引量...
[6] TAIJI SUZUKI.PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additive Model[J].25th Annual Conference on Learning Theory,Workshop and Conference Proceedings,2012,23(8):1-20. [7] MAHDI MILANNI FARD,JOELLE PINEAU.PAC-Bayesian Model Selection forReinforcement Learning[M]....
Herbrich R, Graepel T: A PAC-Bayesian margin bound for lin- ear classifiers: why SVMs work. Proceedings of Advances in Neural Information System Processing 13 (NIPS) 2001:224-230.Herbrich, R., Graepel, T.: A PAC-bayesian margin bound for linear classifiers: Why SVM’s work. Advances ...
Seeger University of Edinburgh University of Edinburgh Overview PAC PAC--Bayesian theorem for Gibbs classifiers Bayesian theorem for Gibbs classifiers Application to Gaussian process Application to Gaussian process classification classification Experiments Experiments Conclusions Conclusions What Is a PAC Bound?
We propose a new theoretical study of domain adaptation for majority vote classifiers (from a source to a target domain). We upper bound the target risk by a trade-off between only two terms: The voters' joint errors on the source domain, and the voters' disagreement on the target one. ...