# PURe: A Plug-and-Play Product-Unit Residual Module for Vision Networks

Source: [PURe: A Plug-and-Play Product-Unit Residual Module for Vision Networks](<https://www.sota2.com/research/paper/68fa787b1e209131cea789c7>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-08-11T05:02:11.191Z

Authors: Ziyuan Li, Uwe Jaekel, Babette Dellen

Published: 2025-05-07T13:21:25.000Z

## Abstract

Modern vision networks are dominated by additive local transformations, whereas explicit multiplicative local interactions remain underexplored. Product units offer a direct approach to modeling such interactions, but their use in deep architectures has been limited by optimization instability. In this work, we propose PURe, a product-unit Residual Module for deep vision networks. PURe is built around a 2D product unit with a real-valued log-domain formulation that makes multiplicative local aggregation practical within deep residual hierarchies. The resulting module serves as a drop-in replacement for native residual units. We instantiate PURe in residual CNNs for image classification and in 2D residual encoder--decoder networks for slice-based segmentation on volumetric CT data. Across Galaxy10 DECaLS, ImageNet, and CIFAR-10, PURe consistently improves residual CNNs and yields a more favorable accuracy--parameter trade-off, allowing moderately deep models to match or surpass substantially deeper ResNet baselines with much smaller parameter budgets. On the AMOS benchmark, PURe also improves slice-based CT segmentation under 3D case-level evaluation. These results show that explicit multiplicative local interaction is a practical and effective design primitive for deep residual vision networks.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Classification on Galaxy10 DECaLS (test)](<https://www.sota2.com/research/sota/classification-on-galaxy10-decals-test.md>): Classification; Galaxy10 DECals (test); Accuracy: 84.28.

- [Classification on Galaxy10 DECaLS Poisson-corrupted (test)](<https://www.sota2.com/research/sota/classification-on-galaxy10-decals-poisson-corrupted-test.md>): Classification; Galaxy10 DECaLS Poisson-corrupted (test); Accuracy: 83.83.

- [Image Classification on ImageNet 1K (val) (Standard Accuracy)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-1k-val-standard-accuracy.md>): Image Classification; ImageNet-1k (val); Top-1 Accuracy: 81.23.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2505.04397>)

[PDF](<https://arxiv.org/pdf/2505.04397.pdf>)
