Mildly Overparameterized ReLU Networks on Orthogonal Data: Incremental Learning and Implicit Bias

Documenti analoghi

• Benignity of loss landscape with weight decay requires both large overparametrization and initialization
  di: Boursier, Etienne, et al.
  Pubblicazione: (2025)
• Simplicity bias and optimization threshold in two-layer ReLU networks
  di: Boursier, Etienne, et al.
  Pubblicazione: (2024)
• Mildly Overparameterized ReLU Networks Have a Favorable Loss Landscape
  di: Karhadkar, Kedar, et al.
  Pubblicazione: (2023)
• Gradient flow dynamics of shallow ReLU networks for square loss and orthogonal inputs
  di: Boursier, Etienne, et al.
  Pubblicazione: (2022)
• Asymptotic Smoothing of the Lipschitz Loss Landscape in Overparameterized One-Hidden-Layer ReLU Networks
  di: Baturin, Saveliy
  Pubblicazione: (2026)