Kronecker-Factored Approximate Curvature for Modern Neural Network Architectures


Loading...

Date

2024-07

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric

Data

Abstract

The core components of many modern neural network architectures, such as transformers, convolutional, or graph neural networks, can be expressed as linear layers with $\textit{weight-sharing}$. Kronecker-Factored Approximate Curvature (K-FAC), a second-order optimisation method, has shown promise to speed up neural network training and thereby reduce computational costs. However, there is currently no framework to apply it to generic architectures, specifically ones with linear weight-sharing layers. In this work, we identify two different settings of linear weight-sharing layers which motivate two flavours of K-FAC -- $\textit{expand}$ and $\textit{reduce}$. We show that they are exact for deep linear networks with weight-sharing in their respective setting. Notably, K-FAC-reduce is generally faster than K-FAC-expand, which we leverage to speed up automatic hyperparameter selection via optimising the marginal likelihood for a Wide ResNet. Finally, we observe little difference between these two K-FAC variations when using them to train both a graph neural network and a vision transformer. However, both variations are able to reach a fixed validation metric target in $50$-$75\%$ of the number of steps of a first-order reference run, which translates into a comparable improvement in wall-clock time. This highlights the potential of applying K-FAC to modern neural network architectures.

Publication status

published

Book title

Advances in Neural Information Processing Systems 36

Journal / series

Volume

Pages / Article No.

33624 - 33655

Publisher

Curran

Event

37th Annual Conference on Neural Information Processing Systems (NeurIPS 2023)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Machine Learning (cs.LG); Machine Learning (stat.ML); FOS: Computer and information sciences; Deep learning; second-order; Optimization; Natural gradient; fisher; Gauss-Newton; k-fac; weight-sharing

Organisational unit

09568 - Rätsch, Gunnar / Rätsch, Gunnar check_circle

Notes

Poster presentation

Funding

Related publications and datasets

Is new version of: 10.48550/ARXIV.2311.00636