Saved in:
Bibliographic Details
Main Authors: Rahman, Md Ashiqur, Hao, Lim Jun, Jiang, Jeremiah, Lim, Teck-Yian, Yeh, Raymond A.
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.26657
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Equivariance is a fundamental property in computer vision models, yet strict equivariance is rarely satisfied in real-world data, which can limit a model's performance. Controlling the degree of equivariance is therefore desirable. We propose a general framework for constructing soft equivariant models by projecting the model weights into a designed subspace. The method applies to any pre-trained architecture and provides theoretical bounds on the induced equivariance error. Empirically, we demonstrate the effectiveness of our method on multiple pre-trained backbones, including ViT and ResNet, across image classification, semantic segmentation, and human-trajectory prediction tasks. Notably, our approach improves the performance while simultaneously reducing equivariance error on the competitive ImageNet benchmark.