Safety
Gradient flow dynamics of shallow ReLU networks for square loss and orthogonal inputs
arXiv:2206.00939v3 Announce Type: replace-cross Abstract: The training of neural networks by gradient descent methods is a cornerstone of the deep learning revolution. Yet, despite some recent progres
arXiv:2206.00939v3 Announce Type: replace-cross Abstract: The training of neural networks by gradient descent methods is a cornerstone of the deep learning revolution. Yet, despite some recent progress, a complete theory explaining its success is still missing. This article presents, for orthogonal input vectors, a precise description of the gradient flow dynamics of training one-hidden layer ReLU neural networks for the mean squared error at small initialisation. In this setting, despite non-convexity, we show that the gradient flow converges to zero loss and characterise its implicit bias towards minimum variation norm. Furthermore, some interesting phenomena are highlighted: a quantitative description of the initial alignment phenomenon and a proof that the process follows a specific saddle to saddle dynamics.
Related
- Optimal Rates for Generalization of Gradient Descent for Deep ReLU Classification
- Implicit Bias in Deep Linear Discriminant Analysis
- Training event-based neural networks with exact gradients via Differentiable ODE Solving in JAX
- Learning Aligned Stability in Neural ODEs Reconciling Accuracy with Robustness
Source: arXiv cs.LG | 2026-04-15