Identifiable Equivariant Networks are Layerwise Equivariant

Vahid Shahverdi, Giovanni Luca Marchetti, Georg Bökman, Kathlén Kohn · 2026

Read Identifiable Equivariant Networks are Layerwise Equivariant by Vahid Shahverdi, Giovanni Luca Marchetti, Georg Bökman, Kathlén Kohn (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-21645.txt.

Description

Vahid Shahverdi, Giovanni Luca Marchetti, Georg Bökman, and Kathlén Kohn prove a tidy fact about symmetry in deep networks. A network is equivariant when shifting the input shifts the output in a matching way. That property can hold for the whole network while the inner layers look messy.

The authors show the mess is optional. For any such network, the same overall function can be realized with layers that are each equivariant too. Layerwise symmetry makes networks easier to analyze and to build by design. The theorem assures designers that imposing symmetry layer by layer costs nothing in what the network can express.

Catalog record

CREATORVahid Shahverdi, Giovanni Luca Marchetti, Georg Bökman, Kathlén Kohn
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-21645

Browse the full library · About the Tower of Babel