arXiv subjects

Su Su

Publications and source records attributed to Su Su.

3 recordsLinked to original sources

arXiv2026

Correct Answers, Closed Gates: SolveBind for GUI Agents at the Verification Barrier

Websites and applications deploy CAPTCHAs and anti-automation mechanisms that use behavioral signals such as interaction trajectories and timing, not outcomes alone, to distinguish humans from automated agents. We call these protected GUI environments, where correctly solving a verification task is insufficient for end-to-end task completion. Agents must turn solver outputs into accepted interactions and adapt execution to runtime feedback under time constraints. We introduce SolveBind, a GUI-agent framework combining hierarchical solving with runtime supervision. A hierarchical solving tree decomposes verification into interdependent Perceive, Ground, Solve, Act, and Verify capabilities. A runtime supervisor grounds action commitments in tool-derived evidence, revises them as evidence arrives, and mediates execution under explicit recovery budgets. Across nine CAPTCHA task families and nine backbone models, SolveBind achieves the highest success rate on eight backbones, outperforming the strongest baseline by 13.3-34.5 percentage points. It achieves the highest success rate on all six AndroidWorld backbones. In a paired GPT-5.5 analysis, MobileAgent produces correct answers but fails dynamic verification on 13 of 90 tasks; SolveBind succeeds on seven of these tasks. With automatic submission, removing binding reduces success from 82.22% to 71.56% and increases latency and token usage. Recovery caps reduce execution cost with a small observed success trade-off: removing them raises success from 82.22% to 83.56%, while increasing latency and token usage to 3.7x and 2.4x their full-supervision values, respectively. These results support evidence-grounded execution control while identifying a success-cost trade-off in recovery and failure boundaries of heuristic submission rules.

cs.AI↗

arXiv2026

In-Context Learning to Assess Built Environment Impacts on Perceived Neighborhood Walkability Among Mobility-impaired Older Adults

As global populations age, enhancing neighborhood walkability through inclusive urban design is important for mitigating built environment (BE) barriers that discourage physical activity and social participation among older adults. This study investigates the utility of in-context learning (ICL), using the transformer-based foundation model TabPFN, to determine how BE features influence perceived walkability, as measured by the Neighborhood Environment Walkability Scale (NEWS-A) survey. Using a small-scale dataset (N = 257) comprising a unique demographic of older adults with knee osteoarthritis or a history of falls, TabPFN achieved a macro F1 score of 54.89% for walkability perceptions categorized as Low, Neutral, and High using equal-width binning. This result outperformed optimized, grid-searched baseline models, including Random Forest (45.85%) and XGBoost (50.56%). To interpret these results, we employed Shapley Interaction Quantification (SHAP-IQ) to identify the hierarchical importance of feature interactions. Preliminary results revealed that the model's predictive logic was primarily driven by higher-order interactions. For example, the interaction between average street circuity and the ratio of drivable roads emerged as the primary discriminator of perceived walkability. Neighborhood greenery was found to have substantial predictive importance only when combined with an individual's fear of falling or perception of age-friendliness. Overall, ICL using TabPFN demonstrates superior performance on small-scale datasets, enhancing the fidelity of the resulting interpretive insights. Furthermore, SHAP-IQ provides a synergistic perspective on how higher-order feature interactions drive the model's predictions.

cs.LG↗

arXiv2026

HLL: Can Agents Cross Humanity's Last Line of Verification?

Multimodal agents are increasingly expected to operate interfaces on behalf of users, raising a central deployment question: can they truly substitute for humans in workflows that services deliberately protect against automation? CAPTCHA verification makes this question concrete. It is not merely a visual puzzle, but a human-verification boundary placed before account creation, content access, form submission, and other protected actions. We introduce \textbf{Humanity's Last Line of Verification (HLL)}, a controlled benchmark that uses interactive CAPTCHA verification to evaluate whether agents can cross this boundary through grounded, human-like interaction rather than recognition alone. HLL covers diverse CAPTCHA interactions and exposes agents to controlled realism stressors, including cluttered webpages, harder task variants, and trace-conditioned validation of the solving process. We evaluate eight frontier multimodal agents in a closed-loop GUI environment. The results show that current agents remain brittle at this human-substitution boundary: performance varies sharply across verification types, degrades under realistic interface conditions, and drops further when correct answers must be supported by valid action traces. By exposing gaps in localization, action calibration, state tracking, and process consistency, HLL provides a concrete testbed for measuring how close multimodal agents are to acting as human substitutes in protected real-world workflows. Our code is available at https://github.com/XinhaoS0101/HLL

cs.AI↗