# Purge-Gate: Backpropagation-Free Test-Time Adaptation for Point Clouds Classification via Token Purging

Source: [Purge-Gate: Backpropagation-Free Test-Time Adaptation for Point Clouds Classification via Token Purging](<https://www.sota2.com/research/paper/68fa7b411e209131cea87531>)

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

Data updated: 2025-10-26T05:35:42.317Z

Authors: Moslem Yazdanpanah, Ali Bahri, Mehrdad Noori, Sahar Dastani, Gustavo Adolfo Vargas Hakim, David Osowiechi, Ismail Ben Ayed, Christian Desrosiers

Published: 2025-09-11T18:33:40.000Z

## Abstract

Test-time adaptation (TTA) is crucial for mitigating performance degradation caused by distribution shifts in 3D point cloud classification. In this work, we introduce Token Purging (PG), a novel backpropagation-free approach that removes tokens highly affected by domain shifts before they reach attention layers. Unlike existing TTA methods, PG operates at the token level, ensuring robust adaptation without iterative updates. We propose two variants: PG-SP, which leverages source statistics, and PG-SF, a fully source-free version relying on CLS-token-driven adaptation. Extensive evaluations on ModelNet40-C, ShapeNet-C, and ScanObjectNN-C demonstrate that PG-SP achieves an average of +10.3\\% higher accuracy than state-of-the-art backpropagation-free methods, while PG-SF sets new benchmarks for source-free adaptation. Moreover, PG is 12.4 times faster and 5.5 times more memory efficient than our baseline, making it suitable for real-world deployment. Code is available at \\hyperlink{https://github.com/MosyMosy/Purge-Gate}{https://github.com/MosyMosy/Purge-Gate}

## 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.

- [3D Point Cloud Classification on ShapeNet-C](<https://www.sota2.com/research/sota/3d-point-cloud-classification-on-shapenet-c.md>): 3D Point Cloud Classification; ShapeNet-C; Accuracy (Uniform Noise): 79.21.

- [3D Point Cloud Classification on ScanObjectNN-C 1.0 (test)](<https://www.sota2.com/research/sota/3d-point-cloud-classification-on-scanobjectnn-c-1-0-test.md>): 3D Point Cloud Classification; ScanObjectNN-C 1.0 (test); Accuracy (Uniform Noise): 41.48.

- [3D Point Cloud Classification on ModelNet40-C (test)](<https://www.sota2.com/research/sota/3d-point-cloud-classification-on-modelnet40-c-test.md>): 3D Point Cloud Classification; ModelNet40-C (test); Robustness to Rotation: 56.94.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2509.09785>)

[PDF](<https://arxiv.org/pdf/2509.09785.pdf>)