AI Paper Atlas

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

Learning from Reliable Negatives: Confidence-Anchored Test-Time Adaptation for GUI Grounding

Yizhou Liu, Fei Tang, Yuchen Yan, Zhengxi Lu, Songqin Nong, Tao Jiang, Wenhao Xu, Wenqi Zhang, Weiming Lu, Jun Xiao, Yongliang Shen

ECCV 2026 · Label-efficient and robust learning › Test-time adaptation · approach: Extrapolation and single-sample adaptation · 1 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
select pseudo-label by coordinate-token confidence instead of all-token confidence (ablation)CAL/CANL · ScreenSpot-V2 pseudo-label (N=12)pseudo-label accuracy: 78.6 -> 83.9Improved
use negative-only learning (CANL) instead of CAL on ScreenSpot-Pro (ablation)Qwen-2.5-VL-7B · ScreenSpot-Proaccuracy: 32.7 -> 33.8Improved
use negative-only learning (CANL) instead of CAL on UI-Vision (ablation)Qwen-2.5-VL-7B · UI-Visionaccuracy: 18.6 -> 20.1Improved
use negative-only learning (CANL) instead of CAL on ScreenSpot-V2 (ablation)Qwen-2.5-VL-7B · ScreenSpot-V2accuracy: 92.1 -> 92.1No clear effect
use negative-only learning (CANL) instead of CAL on AndroidWorld (ablation)Qwen-2.5-VL-3B grounding · AndroidWorldsuccess rate: 30.2 -> 29.3Hurt
negative-only vs positive-only learning (CAPL) (ablation)Qwen-2.5-VL-3B · ScreenSpot-V2accuracy: 86.7 -> 88.5Improved
increase distance threshold tau from 0.05 to 0.1 (CAL) (ablation)CAL · ScreenSpot-V2accuracy: 88.9 -> 86.9Hurt
increase distance threshold tau from 0.05 to 0.1 (CANL) (ablation)CANL · ScreenSpot-V2accuracy: 88.5 -> 88.5No clear effect
use binary reward instead of continuous distance reward (ablation)CAL · ScreenSpot-V2accuracy: 85.2 -> 88.9Improved
use RL (GRPO) instead of SFT on pseudo-labels (ablation)Qwen-2.5-VL-3B · ScreenSpot-V2accuracy: 84.6 -> 88.9Improved
apply label-free CAL/CANL test-time training to base modelQwen-2.5-VL-3B · ScreenSpot-Proaccuracy: 16.1 -> 32.7Improved
apply CAL as grounding in AndroidWorld agentQwen-2.5-VL-3B + GPT-4o planner · AndroidWorldsuccess rate: 25.0 -> 30.2Improved

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