Rethinking precision of pseudo label: Test-time adaptation via complementary learning

Longbin Zeng, Jiayi Han, Liang Du, Weiyang Ding

Abstract: Highlights•The concept of complementary learning is cleverly applied to the problem of test-time adaptation.•This learning framework agrees with Vanilla learning with ordinary data distribution.•Our proposed algorithm exhibits strong generalizability to the field of unsupervised learning.