In high-dimensional linear and one-step feature-learning models, a regularized student can outperform its teacher by fixing under-regularization, using better regularization structure, or retaining pretrained hard features.
The solutions are given by ( λ− s , λ+ s ) from (12)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
On the Mechanisms of Weak-to-Strong Generalization: A Theoretical Perspective
In high-dimensional linear and one-step feature-learning models, a regularized student can outperform its teacher by fixing under-regularization, using better regularization structure, or retaining pretrained hard features.