Jan 9, 2022 · In this paper, we unify and substantially generalize stability based generalization bounds and make three technical contributions.

Recent years have seen advances in generalization bounds for noisy stochastic algorithms, especially stochastic gradient Langevin dynamics (SGLD) based on ...

Jan 28, 2022 · Abstract: We study the generalization of noisy stochastic mini-batch based iterative algorithms based on the notion of stability.

We provide optimization guarantees for two variants of EFLD: ○ For SGLD, result is similar to previous work [Bassily et al.,2014; Wang & Xu, 2019];.

In. Section 3, we introduce Exponential Family Langevin Dynamics (EFLD), a family of noisy gradient descent algorithms based on exponential family noise.

, Stability Based Generalization Bounds for Exponential Family Langevin Dynamics ;; Arindam A Banerjee ;, ISSN: 2640-3498 ;, Proceedings of Machine Learning ...

Jul 19, 2022 · Stability Based Generalization Bounds for Exponential Family Langevin Dynamics. Jul 19, 2022. Speakers. AB · Arindam Banerjee. Speaker · 0 ...

Connected Papers is a visual tool to help researchers and applied scientists find academic papers relevant to their field of work.

Sep 13, 2024 · Stability Based Generalization Bounds for Exponential Family Langevin Dynamics. ICML 2022: 1412-1449. [c7]. view. electronic edition via DOI ...

Stability Based Generalization Bounds for Exponential Family Langevin Dynamics ... Hessian based analysis of SGD for Deep Nets: Dynamics and Generalization