all AI news
RDumb: A simple approach that questions our progress in continual test-time adaptation
April 4, 2024, 4:42 a.m. | Ori Press, Steffen Schneider, Matthias K\"ummerer, Matthias Bethge
cs.LG updates on arXiv.org arxiv.org
Abstract: Test-Time Adaptation (TTA) allows to update pre-trained models to changing data distributions at deployment time. While early work tested these algorithms for individual fixed distribution shifts, recent work proposed and applied methods for continual adaptation over long timescales. To examine the reported progress in the field, we propose the Continually Changing Corruptions (CCC) benchmark to measure asymptotic performance of TTA techniques. We find that eventually all but one state-of-the-art methods collapse and perform worse than …
abstract algorithms arxiv continual cs.cv cs.lg data deployment distribution pre-trained models progress questions simple test type update work
More from arxiv.org / cs.LG updates on arXiv.org
Jobs in AI, ML, Big Data
Founding AI Engineer, Agents
@ Occam AI | New York
AI Engineer Intern, Agents
@ Occam AI | US
AI Research Scientist
@ Vara | Berlin, Germany and Remote
Data Architect
@ University of Texas at Austin | Austin, TX
Data ETL Engineer
@ University of Texas at Austin | Austin, TX
Sr. BI Analyst
@ AkzoNobel | Pune, IN