Stability and Deviation Optimal Risk Bounds with Convergence Rate O(1/n)


Loading...

Date

2021

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric

Data

Abstract

The sharpest known high probability generalization bounds for uniformly stable algorithms (Feldman, Vondrak, NeurIPS 2018, COLT, 2019), (Bousquet, Klochkov, Zhivotovskiy, COLT, 2020) contain a generally inevitable sampling error term of order Θ(1/√n). When applied to excess risk bounds, this leads to suboptimal results in several standard stochastic convex optimization problems. We show that if the so-called Bernstein condition is satisfied, the term Θ(1/√n) can be avoided, and high probability excess risk bounds of order up to O(1/n) are possible via uniform stability. Using this result, we show a high probability excess risk bound with the rate O(logn/n) for strongly convex and Lipschitz losses valid for \emph{any} empirical risk minimization method. This resolves a question of Shalev-Shwartz, Shamir, Srebro, and Sridharan (COLT, 2009). We discuss how O(logn/n) high probability excess risk bounds are possible for projected gradient descent in the case of strongly convex and Lipschitz losses without the usual smoothness assumption.

Publication status

published

External links

Book title

Advances in Neural Information Processing Systems 34 (NeurIPS 2021)

Volume

31

Pages / Article No.

5065 - 5076

Publisher

Curran

Event

35th Conference on Neural Information Processing Systems (NeurIPS 2021)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Organisational unit

09679 - Bandeira, Afonso / Bandeira, Afonso check_circle

Notes

Funding

Related publications and datasets