For any Gaussian multi-index model, the generative leap exponent k⋆ sharply characterizes the sample complexity of efficient subspace recovery as Θ(d^(1∨k⋆/2)).
Repetita iuvant: Data repetition allows sgd to learn high-dimensional multi-index functions, 2024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Generative Leap: Sharp Sample Complexity for Efficiently Learning Gaussian Multi-Index Models
For any Gaussian multi-index model, the generative leap exponent k⋆ sharply characterizes the sample complexity of efficient subspace recovery as Θ(d^(1∨k⋆/2)).