AI RESEARCH
T-QPM: Enabling Temporal Out-Of-Distribution Detection and Domain Generalization for Vision-Language Models in Open-World
arXiv CS.LG
•
ArXi:2603.18481v1 Announce Type: cross Out-of-distribution (OOD) detection remains a critical challenge in open-world learning, where models must adapt to evolving data distributions. While recent vision-language models (VLMS) like CLIP enable multimodal OOD detection through Dual-Pattern Matching (DPM), existing methods typically suffer from two major shortcomings: (1) They rely on fixed fusion rules and assume static environments, failing under temporal drift; and (2) they lack robustness against covariate shifted inputs.