Neural LoFi models deep learning as layer-wise spectral filtering that selects maximal low-degree correlations, yielding a tractable surrogate for hierarchical representation learning beyond the lazy regime.
Emergence of simple-cell receptive field properties by learning a sparse code for natural images.Nature, 381(6583):607–609
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
DOODL learns a dictionary of spectral dynamics to approximate a manifold of related dynamical systems, enabling compact representations and improved operator estimation from short or partial trajectories.
Strong superposition causes neural loss to scale as the inverse of model dimension due to geometric feature overlaps, explaining scaling laws for broad frequency distributions.
citing papers explorer
-
Deep Learning as Neural Low-Degree Filtering: A Spectral Theory of Hierarchical Feature Learning
Neural LoFi models deep learning as layer-wise spectral filtering that selects maximal low-degree correlations, yielding a tractable surrogate for hierarchical representation learning beyond the lazy regime.
-
Geometric Dictionary Learning of Dynamical Systems with Optimal Transport
DOODL learns a dictionary of spectral dynamics to approximate a manifold of related dynamical systems, enabling compact representations and improved operator estimation from short or partial trajectories.
-
Superposition Yields Robust Neural Scaling
Strong superposition causes neural loss to scale as the inverse of model dimension due to geometric feature overlaps, explaining scaling laws for broad frequency distributions.