AI TrackerFrontier desk · AEST
Back to news
normalaitech

Neural Expansion: A Unified Mechanism for How Deep Neural Networks Generalize

arXiv cs.AI · 17 Oct 2026, 02:38 AEST

A new framework called Neural Expansion explains how deep neural networks generalize despite being over-parameterized. It characterizes generalization through the expansion of directionally-selective ray coverage in the input space.

Key facts

Read aloud

A new framework called Neural Expansion explains how deep neural networks generalize despite being over-parameterized. It characterizes generalization through the expansion of directionally-selective ray coverage in the input space. Deep neural networks demonstrate remarkable generalization despite being highly over-parameterized, challenging traditional statistical learning theories. We introduce Neural Expansion, a unified geometric framework that characterizes generalization through the expansion of directionally-selective ray coverage in the input space. Using Generalization Intervals GIs, we quantify direction-specific regions around training samples where the model's predictions remain stable. Our extensive empirical studies across diverse architectures and datasets reveal that training leads to progressive expansion of ray coverage, particularly along directions with low Jacobian singular values.

Sources in this cluster