A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is
Test-time adaptation (TTA) has increasingly been an important topic to efficiently tackle the cross-domain distribution shift at test time for medical images from different institutions. ... To address this problem, we propose a novel dynamic learning rate adjustment method for test-time adaptation, called DLTTA, which dynamically modulates the amount of weights update for each test image ... Our main contributions are as follows: • We propose a new dynamic learning rate adjustment framework for test-time adaptation on cross-domain medical image analysis. ...arXiv:2205.13723v1 fatcat:udwtqatuc5ef5pisq3qccvropi