Test-Time Training with Masked Autoencoders
NeurIPS 2022(2022)
摘要
Test-time training adapts to a new test distribution on the fly by optimizing a model for each test input using self-supervision. In this paper, we use masked autoencoders for this one-sample learning problem. Empirically, our simple method improves generalization on many visual benchmarks for distribution shifts. Theoretically, we characterize this improvement in terms of the bias-variance trade-off.
更多查看译文
关键词
Computer Vision,Test-Time Training,Masked Auto-Encoder
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要