Keywords: equivariance, so(3) symmetry, tensor data, physics
TL;DR: We design a modular, equivariant neural architecture that generalizes affine layers to SO(3) representations and exploits expressive bilinear operations, to improve learning on scalar, vector, and tensor-valued data.
Abstract: Many datasets in scientific and engineering applications are comprised of objects which have
specific geometric structure. A common example is data which inhabits a representation of the
group SO(3) of 3D rotations: scalars, vectors, tensors, etc. One way for a neural network to
exploit prior knowledge of this structure is to enforce SO(3)-equivariance throughout its layers, and
several such architectures have been proposed. While general methods for handling arbitrary SO(3)
representations exist, they computationally intensive and complicated to implement. We show that
by judicious symmetry breaking, we can efficiently increase the expressiveness of a network operating
only on vector and order-2 tensor representations of SO(2). We demonstrate the method on an
important problem from High Energy Physics known as b-tagging, where particle jets originating
from b-meson decays must be discriminated from an overwhelming QCD background. In this task,
we find that augmenting a standard architecture with our method results in a 2.3× improvement in
rejection score.
Submission Number: 15315
Loading