L2M: Practical posterior Laplace approximation with optimization-driven second moment estimation
arXiv:2107.04695
Abstract
Uncertainty quantification for deep neural networks has recently evolved through many techniques. In this work, we revisit Laplace approximation, a classical approach for posterior approximation that is computationally attractive. However, instead of computing the curvature matrix, we show that, under some regularity conditions, the Laplace approximation can be easily constructed using the gradient second moment. This quantity is already estimated by many exponential moving average variants of Adagrad such as Adam and RMSprop, but is traditionally discarded after training. We show that our method (L2M) does not require changes in models or optimization, can be implemented in a few lines of code to yield reasonable results, and it does not require any extra computational steps besides what is already being computed by optimizers, without introducing any new hyperparameter. We hope our method can open new research directions on using quantities already computed by optimizers for uncertainty estimation in deep neural networks.
arXiv comment6 pages, 1 figure, accepted for ICML 2021 UDL Workshop
References in corpus (9)
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Decoupled Weight Decay Regularization
- Overcoming catastrophic forgetting in neural networks
- ADADELTA: An Adaptive Learning Rate Method
- On the Convergence of Adam and Beyond
- On Empirical Comparisons of Optimizers for Deep Learning
- Limitations of the Empirical Fisher Approximation for Natural Gradient Descent
- Scaling Hamiltonian Monte Carlo Inference for Bayesian Neural Networks with Symmetric Splitting
- On the interplay between noise and curvature and its effect on optimization and generalization