ADLER -- An efficient Hessian-based strategy for adaptive learning rate

no code implementations • 25 May 2023 • Dario Balboni, Davide Bacciu

We also compare the novel approximation with the Gauss-Newton approximation.

A partial theory of Wide Neural Networks using WC functions and its practical implications

no code implementations • 29 Sep 2021 • Dario Balboni, Davide Bacciu

We present a framework based on the theory of Polyak-Łojasiewicz functions to explain the properties of convergence and generalization of overparameterized feed-forward neural networks.