Input-gradient space particle inference for neural network ensembles
No Thumbnail Available
Access rights
openAccess
publishedVersion
URL
Journal Title
Journal ISSN
Volume Title
A4 Artikkeli konferenssijulkaisussa
This publication is imported from Aalto University research portal.
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Other link related to publication (opens in new window)
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Other link related to publication (opens in new window)
Date
2024
Department
Major/Subject
Mcode
Degree programme
Language
en
Pages
19
Series
12th International Conference on Learning Representations (ICLR 2024)
Abstract
Deep Ensembles (DEs) demonstrate improved accuracy, calibration and robustness to perturbations over single neural networks partly due to their functional diversity. Particle-based variational inference (ParVI) methods enhance diversity by formalizing a repulsion term based on a network similarity kernel. However, weight-space repulsion is inefficient due to over-parameterization, while direct function-space repulsion has been found to produce little improvement over DEs. To sidestep these difficulties, we propose First-order Repulsive Deep Ensemble (FoRDE), an ensemble learning method based on ParVI, which performs repulsion in the space of first-order input gradients. As input gradients uniquely characterize a function up to translation and are much smaller in dimension than the weights, this method guarantees that ensemble members are functionally different. Intuitively, diversifying the input gradients encourages each network to learn different features, which is expected to improve the robustness of an ensemble. Experiments on image classification datasets and transfer learning tasks show that FoRDE significantly outperforms the gold-standard DEs and other ensemble methods in accuracy and calibration under covariate shift due to input perturbations.Description
| openaire: EC/H2020/951847/EU//ELISE
Keywords
Other note
Citation
Trinh, T, Heinonen, M, Acerbi, L & Kaski, S 2024, Input-gradient space particle inference for neural network ensembles . in 12th International Conference on Learning Representations (ICLR 2024) . Curran Associates Inc., International Conference on Learning Representations, Vienna, Austria, 07/05/2024 . < https://openreview.net/forum?id=nLWiR5P3wr >