Extends random matrix theory via High-dimensional Equivalents to characterize training and generalization performance of linear models, shallow networks, and deep networks in the proportional high-dimensional regime.
Kernel spectral clustering of large dimensional data,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ML 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Random Matrix Theory for Deep Learning: Beyond Eigenvalues of Linear Models
Extends random matrix theory via High-dimensional Equivalents to characterize training and generalization performance of linear models, shallow networks, and deep networks in the proportional high-dimensional regime.