A Variant of Azuma's Inequality for Martingales with Subgaussian Tails
arXiv:1110.2392
Abstract
We provide a variant of Azuma's concentration inequality for martingales, in which the standard boundedness requirement is replaced by the milder requirement of a subgaussian tail.
Cited by in corpus (10)
- A Convergence Theory for Deep Learning via Over-Parameterization
- Spectral bandits for smooth graph functions
- Estimating Structured Vector Autoregressive Model
- Efficient active learning of sparse halfspaces with arbitrary bounded noise
- Learning a Single Neuron with Gradient Methods
- Faster Convergence of Stochastic Gradient Langevin Dynamics for Non-Log-Concave Sampling
- Learning Mixture of Gaussians with Streaming Data
- Efficient and Consistent Robust Time Series Analysis
- Understanding Modern Techniques in Optimization: Frank-Wolfe, Nesterov's Momentum, and Polyak's Momentum
- Regularization by Misclassification in ReLU Neural Networks