Timezone: »

 
2 CENTs on continual adaptation: replay & parameter buffers stabilize entropy minimization
Ori Press · Steffen Schneider · Matthias Kuemmerer · Matthias Bethge

We propose continual entropy minimization (CENT) to adapt computer vision models to continual distribution shifts at ImageNet scale. CENT leverages a replay buffer with images from the source distribution along with rolling parameter buffers to stabilize the training dynamics of conventional test-time adaptation methods. Our work is the first to demonstrate stable, continual adaptation on ImageNet scale, and obtains state-of-the-art results in both static and continual variants of the ImageNet-C benchmark.

Author Information

Ori Press
Steffen Schneider (University of Tuebingen / EPFL / ELLIS)
Matthias Kuemmerer (Center for Integrative Neuroscience, University of Tübingen)
Matthias Bethge (University of Tübingen)

More from the Same Authors