AI Paper Atlas

Research map / Label-efficient and robust learning / Test-time adaptation

Neutralizing Token Aggregation via Information Augmentation for Efficient Test-Time Adaptation

Yizhe Xiong, Zihan Zhou, Yiwen Liang, Hui Chen, Zijia Lin, Xinhao Xu, Tianxiang Hao, Fan Zhang, Jungong Han, Guiguang Ding

ECCV 2026 · Label-efficient and robust learning › Test-time adaptation · approach: Realistic and stable test-time adaptation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add [CLS] embedding augmentation (ablation)NAVIA r=4 · four OOD benchmarksavg accuracy (%): 66.3 -> 66.6Improved
add shallow layer [CLS] bias augmentation (ablation)NAVIA r=4 · four OOD benchmarksavg accuracy (%): 66.3 -> 66.8Improved
combine both augmentations (ablation)NAVIA r=4 · four OOD benchmarksavg accuracy (%): 66.3 -> 67.2Improved
add [CLS] embedding augmentation (ablation)NAVIA r=8 · four OOD benchmarksavg accuracy (%): 65.6 -> 66.2Improved
add shallow layer [CLS] bias augmentation (ablation)NAVIA r=8 · four OOD benchmarksavg accuracy (%): 65.6 -> 66.1Improved
combine both augmentations (ablation)NAVIA r=8 · four OOD benchmarksavg accuracy (%): 65.6 -> 66.7Improved
augment shallow instead of deep layers (ablation)NAVIA · ImageNet-Cavg accuracy (%): 69.2 -> 69.8Improved
augment shallow instead of uniform layers (ablation)NAVIA · ImageNet-Cavg accuracy (%): 69.2 -> 69.8Improved
tune pre-trained [CLS] instead of newly added tokens (ablation)NAVIA · ImageNet-Cavg accuracy (%): 69.4 -> 69.8Improved
tune pre-trained [CLS] instead of re-initialized [CLS] (ablation)NAVIA · ImageNet-Cavg accuracy (%): 65.9 -> 69.8Improved
increase token reduction from r=4 to r=8NAVIA · ImageNet-Cavg accuracy (%): 70.7 -> 69.8Hurt
use NAVIA_4 instead of ToMe_4ToMe · ImageNet-Cavg accuracy (%): 69.9 -> 70.7Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy