Proves linear max-information bound for DP-SGD and derives explicit PAC-Bayes generalization bounds with learnable prior and hyperparameter-controlled complexity term.
Some theoretical improvements on the tightness of PAC-Bayes risk certificates for neural networks.arXiv preprint arXiv:2510.07935,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
From Privacy to Generalization: Linear Max-Information Bounds for DP-SGD
Proves linear max-information bound for DP-SGD and derives explicit PAC-Bayes generalization bounds with learnable prior and hyperparameter-controlled complexity term.