The Spectral Dimension of NTKs is Constant: A Theory of Implicit Regularization, Finite-Width Stability, and Scalable Estimation

Documenti analoghi

• Implicit Regularization in Perturbed Deep Matrix Factorization: Spectral Conditions and Stability
  di: Wang, Jingzhe, et al.
  Pubblicazione: (2026)
• Some Theoretical Results on Layerwise Effective Dimension Oscillations in Finite Width ReLU Networks
  di: Makwana, Darshan
  Pubblicazione: (2025)
• Estimating Implicit Regularization in Deep Learning
  di: Rudoler, Joseph H., et al.
  Pubblicazione: (2026)
• Local Hessian Spectral Filtering for Robust Intrinsic Dimension Estimation
  di: Osada, Genki
  Pubblicazione: (2026)
• Efficient Techniques for Data Reconstruction, with Finite-Width Recovery Guarantees
  di: Tansley, Edward, et al.
  Pubblicazione: (2026)