%0 Journal Article %1 rivasplata2019pacbayes %A Rivasplata, Omar %A Tankasali, Vikram M %A Szepesvari, Csaba %D 2019 %K bayesian bounds generalization readings %T PAC-Bayes with Backprop %U http://arxiv.org/abs/1908.07380 %X We explore the family of methods "PAC-Bayes with Backprop" (PBB) to train probabilistic neural networks by minimizing PAC-Bayes bounds. We present two training objectives, one derived from a previously known PAC-Bayes bound, and a second one derived from a novel PAC-Bayes bound. Both training objectives are evaluated on MNIST and on various UCI data sets. Our experiments show two striking observations: we obtain competitive test set error estimates (~1.4% on MNIST) and at the same time we compute non-vacuous bounds with much tighter values (~2.3% on MNIST) than previous results. These observations suggest that neural nets trained by PBB may lead to self-bounding learning, where the available data can be used to simultaneously learn a predictor and certify its risk, with no need to follow a data-splitting protocol.

@article{rivasplata2019pacbayes, abstract = {We explore the family of methods "PAC-Bayes with Backprop" (PBB) to train probabilistic neural networks by minimizing PAC-Bayes bounds. We present two training objectives, one derived from a previously known PAC-Bayes bound, and a second one derived from a novel PAC-Bayes bound. Both training objectives are evaluated on MNIST and on various UCI data sets. Our experiments show two striking observations: we obtain competitive test set error estimates (~1.4% on MNIST) and at the same time we compute non-vacuous bounds with much tighter values (~2.3% on MNIST) than previous results. These observations suggest that neural nets trained by PBB may lead to self-bounding learning, where the available data can be used to simultaneously learn a predictor and certify its risk, with no need to follow a data-splitting protocol.}, added-at = {2019-12-22T19:49:46.000+0100}, author = {Rivasplata, Omar and Tankasali, Vikram M and Szepesvari, Csaba}, biburl = {https://www.bibsonomy.org/bibtex/2c8969427b4ee3f014ecb06f97f68573f/kirk86}, description = {[1908.07380] PAC-Bayes with Backprop}, interhash = {8f392d1aeb07b95cb2c9d2eec6953ff0}, intrahash = {c8969427b4ee3f014ecb06f97f68573f}, keywords = {bayesian bounds generalization readings}, note = {cite arxiv:1908.07380}, timestamp = {2019-12-22T19:49:46.000+0100}, title = {PAC-Bayes with Backprop}, url = {http://arxiv.org/abs/1908.07380}, year = 2019 }