Metraipolis › archive › dc_ae5d9322ca07
Learning from Reliable Negatives: Confidence-Anchored Test-Time Adaptation for GUI Grounding — arXiv:2609.15307
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.15307
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T01:53:27.464Z
- SHA-256 of the archived text
ef3f95435a8f48e5f4dbe4d622f4438df29f12b13081620a79ea030703cf233c- Stable URL of this page
https://metraipolis.com/documento/dc_ae5d9322ca07
How to cite
Learning from Reliable Negatives: Confidence-Anchored Test-Time Adaptation for GUI Grounding — arXiv:2609.15307 — arXiv (en), https://arxiv.org/abs/2609.15307 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ae5d9322ca07 (Metraipolis archive dc_ae5d9322ca07, sha256 ef3f95435a8f48e5f4dbe4d622f4438df29f12b13081620a79ea030703cf233c)
BibTeX
@misc{dcae5d9322ca07,
title = {Learning from Reliable Negatives: Confidence-Anchored Test-Time Adaptation for GUI Grounding — arXiv:2609.15307},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.15307},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ae5d9322ca07; sha256 ef3f95435a8f48e5f4dbe4d622f4438df29f12b13081620a79ea030703cf233c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.15307. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Graphical User Interface (GUI) grounding is essential for autonomous agents to map natural language instructions to precise screen coordinates. However, existing supervised fine-tuning and reinforcement learning methods are constrained by the high cost of annotation, creating a scalability bottleneck. In this paper, we introduce a label-free test-time training paradigm driven by two key insights: (1) confidence patterns in coordinate tokens are a better indicator than full-sequence confidence, and (2) in sparse GUI coordinate spaces, negative samples offer more reliable learning signals than potentially noisy positive ones. We first propose Confidence-Anchored Learning (CAL), which utilizes coordinate-token confidence to filter pseudo-labels and assign distance-based binary rewards. Building on this, we develop Confidence-Anchored Negative Learning (CANL), which exclusively optimizes the model using negative samples to bypass the risks of incorrect positive samples. Experimental results demonstrate that CANL-7B achieves 92.1% on ScreenSpot-V2. On more challenging ScreenSpot-Pro, CANL-7B reaches 33.8%, an 8.9% absolute improvement over the base model. Our findings establish coordinate-token confidence as a powerful alternative to manual annotations for scalable GUI agent development. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-15 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.15307 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T01:25:23.663Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).