Hybrid Random Features
- K. Choromanski, Haoxian Chen, Adrian Weller
- Computer ScienceInternational Conference on Learning…
- 8 October 2021
We propose a new class of random feature methods for linearizing softmax and Gaussian kernels called hybrid random features (HRFs) that automatically adapt the quality of kernel estimation to provide…
Minimax Optimal Estimation of Stability Under Distribution Shift
- Hongseok Namkoong, Yuanzhe Ma, P. Glynn
- Computer ScienceArXiv
- 13 December 2022
The practical utility of the proposed stability framework is demonstrated by using it to compare system designs on problems where robustness to distribution shift is critical and shows that evaluating stability against large performance degradation incurs a statistical cost.