AI Paper Atlas

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

Tempora: Characterising the Time-Contingent Utility of Online Test-Time Adaptation

Sudarshan Sreeram, Young D. Kwon, Cecilia Mascolo

ICML 2026 · Label-efficient and robust learning › Test-time adaptation · approach: Pseudo-label test-time adaptation · 0 citations (OpenAlex)

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 test-time adaptation with AdaBN to standard inference under hard deadlinesStandard ResNet-50 · ImageNet-C rho=100%discrete utility (%): 18.2 -> 30.8Improved
use gradient-based ETA instead of gradient-free AdaBNETA · ImageNet-C rho=100%discrete utility (%): 30.8 -> 18.7Hurt
use SAR instead of standard inference under hard deadlinesSAR · ImageNet-C rho=100%discrete utility (%): 18.2 -> 7.8Hurt
use ETA instead of AdaBN under continuous latency penaltyETA · ImageNet-C T=50mscontinuous utility (%): 28.42 -> 7.22Hurt
use gradient-based Tent under amortised budget B=1s instead of standardTent · ImageNet-C B=1samortised utility (%): 18.16 -> 0.84Hurt
use SHOT-IM instead of other gradient-based methods under amortised budgetSHOT-IM · ImageNet-C B=1samortised utility (%): 0.87 -> 32.22Improved
relax temporal pressure from rho=100% to rho=25% for rank agreement with offlineoffline ranking · ImageNet-C discretemean Spearman correlation: -0.19 -> 0.97Improved
choose CMF (best offline) under temporal evaluationCMF · ImageNet-C 240 evaluationstimes ranked first: 10/15 offline wins -> 29 of 240 (12.1%) temporal winsHurt
choose ETA (no offline wins) under temporal evaluationETA · ImageNet-C 240 evaluationstimes ranked first: 0/15 offline wins -> 103 of 240 (42.9%)Improved
use LAME (designed for non-i.i.d. streams) in i.i.d. episodic evaluationLAME · ImageNet-C 240 evaluationsaccuracy vs standardHurt

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