On the Double Descent of Random Features Models Trained with SGD
–Neural Information Processing Systems
We study generalization properties of random features (RF) regression in high dimensions optimized by stochastic gradient descent (SGD) in under-/over-parameterized regime.
Neural Information Processing Systems
Aug-19-2025, 13:26:52 GMT
- Country:
- Europe
- Belgium > Flanders
- Flemish Brabant > Leuven (0.05)
- United Kingdom > England
- Cambridgeshire > Cambridge (0.04)
- Belgium > Flanders
- North America > United States (0.14)
- Europe
- Genre:
- Research Report > New Finding (0.69)
- Technology: