SOURCE-LINKED INTELLIGENCE

Stability and Generalization of Straight-Through Estimators for Training Two-Layer Quantized Neural Networks

arXiv · AI, language, vision and robotics · article · Sep 6, 2026 · UTC

We study the identity straight-through estimator (STE) for training a two-layer binary-activation network with hinge loss from the perspective of Statistical Learning Theory (SLT). Our central question is whether algorithmic stability can explain the statistical generalization of the estimator produced by the discontinuous STE training rule. In the saturated-output regime, the zero-initialized samplewise STE recursion is exactly the stochastic subgradient descent on the convex latent loss $(-yu^\top x)_+$. This representation makes a stability analysis possible. We derive an exact distance ide

Read original source ↗ Open in workspace

recordType

paper

region

Global