2 papers
stat.ML2025
On Rademacher Complexity-based Generalization Bounds for Deep Learning
Lan V. Truong
We show that the Rademacher complexity-based framework can establish non-vacuous generalization bounds for Convolutional Neural Networks (CNNs) in the context of classifying a smal…
stat.ML2025
Global Convergence Rate of Deep Equilibrium Models with General Activations
Lan V. Truong
In a recent paper, Ling et al. investigated the over-parametrized Deep Equilibrium Model (DEQ) with ReLU activation. They proved that the gradient descent converges to a globally o…