Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Accelerated Linearized Laplace Approximation for Bayesian Deep Learning

About

Laplace approximation (LA) and its linearized variant (LLA) enable effortless adaptation of pretrained deep neural networks to Bayesian neural networks. The generalized Gauss-Newton (GGN) approximation is typically introduced to improve their tractability. However, LA and LLA are still confronted with non-trivial inefficiency issues and should rely on Kronecker-factored, diagonal, or even last-layer approximate GGN matrices in practical use. These approximations are likely to harm the fidelity of learning outcomes. To tackle this issue, inspired by the connections between LLA and neural tangent kernels (NTKs), we develop a Nystrom approximation to NTKs to accelerate LLA. Our method benefits from the capability of popular deep learning libraries for forward mode automatic differentiation, and enjoys reassuring theoretical guarantees. Extensive studies reflect the merits of the proposed method in aspects of both scalability and performance. Our method can even scale up to architectures like vision transformers. We also offer valuable ablation studies to diagnose our method. Code is available at \url{https://github.com/thudzj/ELLA}.

Zhijie Deng, Feng Zhou, Jun Zhu• 2022

Related benchmarks

TaskDatasetResultRank
Image ClassificationFashionMNIST (test)
Accuracy86.6
461
Image ClassificationImageNet (test)--
235
Image ClassificationCIFAR-10 (test)
Accuracy94.4
44
Out-of-Distribution DetectionKMNIST (test)
AUROC0.783
22
RegressionTaxi (test)
NLL3.753
19
RegressionYear (test)
NLL3.674
19
Out-of-Distribution DetectionFashionMNIST (In) / MNIST (Out) (test)
AUROC0.874
18
Image ClassificationImageNet
NLL0.877
16
RegressionAirline (test)
NLL5.086
14
Image ClassificationMNIST (test)
Accuracy97.6
9
Showing 10 of 12 rows

Other info

Follow for update