AI Paper Atlas

Research map / Continual learning / Domain and prompt-based

Decomposing the Neurons: Activation Sparsity via Mixture of Experts for Continual Test Time Adaptation

Rongyu Zhang, Aosong Cheng, Yulin Luo, Gaole Dai, Huanrui Yang, Jiaming Liu, Ran Xu, Li Du, Dan Wang, Yuan Du

AAAI 2026 · Continual learning › Domain and prompt-based · approach: Continual 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 plain 4-expert MoE (ablation)CoTTA baseline · Cityscapes-to-ACDCmean mIoU: 58.6 -> 57.9Hurt
add Spatial Differentiable Dropout (SDD) to experts (ablation)MoE + CoTTA · Cityscapes-to-ACDCmean mIoU: 57.9 -> 61.1Improved
add Domain-Aware Router (DAR) (ablation)MoE + SDD · Cityscapes-to-ACDCmean mIoU: 61.1 -> 61.5Improved
add Activation Sparsity Gate (ASG) (ablation)MoE + SDD + DAR · Cityscapes-to-ACDCmean mIoU: 61.5 -> 62.0Improved
add Homeostatic-Proximal (HP) loss (ablation)MoE + SDD + DAR + ASG · Cityscapes-to-ACDCmean mIoU: 62.0 -> 62.2Improved
use 4 experts instead of 2/8/16 (ablation)MoASE · Cityscapes-to-ACDC round 1mean mIoU: E=2 61.5, E=4 61.8, E=8 60.5, E=16 61.2Mixed
MoASE vs ViDA on ImageNet-CViDA · ImageNet-to-ImageNet-Cmean error (%): 43.4 -> 40.5Improved
MoASE vs ViDA on CIFAR10-CViDA · CIFAR10-to-CIFAR10-Cmean error (%): 20.7 -> 16.8Improved
MoASE vs ViDA on Cityscapes-to-ACDCViDA · Cityscapes-to-ACDC (Segformer-B5)mean mIoU: 61.9 -> 62.2Improved
apply MoASE to Segformer-B0 and SAM-SETRViDA · Cityscapes-to-ACDCmean mIoU: 47.6 -> 48.2 (B0); 65.6 -> 66.0 (SAM-SETR)Improved
MoASE on domain generalization to 5 unseen ImageNet-C domainsViDA · ImageNet-C DGmean error (%): 42.3 -> 37.5Improved
repeat target domain sequence for 3 roundsMoASE · Cityscapes-to-ACDCmean mIoU: 61.8 -> 62.3 -> 62.3Improved

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