 On the Role of Noise in the Sample Complexity of Learning Recurrent
Neural Networks: Exponential Gaps for Long SequencesNeural Information Processing Systems (NeurIPS), 2023 |
 Limitations of Information-Theoretic Generalization Bounds for Gradient
Descent Methods in Stochastic Convex OptimizationInternational Conference on Algorithmic Learning Theory (ALT), 2022 |
