AI RESEARCH
ANCHOR: Branch-Point Data Generation for GUI Agents
arXiv CS.AI
•
ArXi:2602.07153v2 Announce Type: replace End-to-end GUI agents for real desktop environments require large amounts of high-quality interaction data, yet collecting human nstrations is expensive and existing synthetic pipelines often suffer from limited task diversity or noisy, goal-drifting trajectories. We present a trajectory expansion framework Anchor that bootstraps scalable desktop supervision from a small set of verified seed nstrations.