Emergent Equivariance in Deep Ensembles

JE Gerken, P Kessel - arxiv preprint arxiv:2403.03103, 2024 - arxiv.org
We demonstrate that deep ensembles are secretly equivariant models. More precisely, we
show that deep ensembles become equivariant for all inputs and at all training times by …

Ensembles provably learn equivariance through data augmentation

O Nordenfors, A Flinth - arxiv preprint arxiv:2410.01452, 2024 - arxiv.org
Recently, it was proved that group equivariance emerges in ensembles of neural networks
as the result of full augmentation in the limit of infinitely wide neural networks (neural tangent …

Symmetries in Overparametrized Neural Networks: A Mean-Field View

J Maass, J Fontbona - arxiv preprint arxiv:2405.19995, 2024 - arxiv.org
We develop a Mean-Field (MF) view of the learning dynamics of overparametrized Artificial
Neural Networks (NN) under data symmetric in law wrt the action of a general compact …