Unraveling Batch Normalization for Realistic Test-Time Adaptation

Authors

  • Zixian Su School of Advanced Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China Faculty of Science and Engineering, University of Liverpool, Liverpool, the United Kingdom
  • Jingwei Guo School of Advanced Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China Faculty of Science and Engineering, University of Liverpool, Liverpool, the United Kingdom
  • Kai Yao School of Advanced Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China Faculty of Science and Engineering, University of Liverpool, Liverpool, the United Kingdom
  • Xi Yang School of Advanced Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China
  • Qiufeng Wang School of Advanced Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China
  • Kaizhu Huang Data Science Research Center, Duke Kunshan University, Kunshan, China

DOI:

https://doi.org/10.1609/aaai.v38i13.29436

Keywords:

ML: Transfer, Domain Adaptation, Multi-Task Learning, ML: Classification and Regression

Abstract

While recent test-time adaptations exhibit efficacy by adjusting batch normalization to narrow domain disparities, their effectiveness diminishes with realistic mini-batches due to inaccurate target estimation. As previous attempts merely introduce source statistics to mitigate this issue, the fundamental problem of inaccurate target estimation still persists, leaving the intrinsic test-time domain shifts unresolved. This paper delves into the problem of mini-batch degradation. By unraveling batch normalization, we discover that the inexact target statistics largely stem from the substantially reduced class diversity in batch. Drawing upon this insight, we introduce a straightforward tool, Test-time Exponential Moving Average (TEMA), to bridge the class diversity gap between training and testing batches. Importantly, our TEMA adaptively extends the scope of typical methods beyond the current batch to incorporate a diverse set of class information, which in turn boosts an accurate target estimation. Built upon this foundation, we further design a novel layer-wise rectification strategy to consistently promote test-time performance. Our proposed method enjoys a unique advantage as it requires neither training nor tuning parameters, offering a truly hassle-free solution. It significantly enhances model robustness against shifted domains and maintains resilience in diverse real-world scenarios with various batch sizes, achieving state-of-the-art performance on several major benchmarks. Code is available at https://github.com/kiwi12138/RealisticTTA.
AAAI-24 / IAAI-24 / EAAI-24 Proceedings Cover

Downloads

Published

2024-03-24

How to Cite

Su, Z., Guo, J., Yao, K., Yang, X., Wang, Q., & Huang, K. (2024). Unraveling Batch Normalization for Realistic Test-Time Adaptation. Proceedings of the AAAI Conference on Artificial Intelligence, 38(13), 15136-15144. https://doi.org/10.1609/aaai.v38i13.29436

Issue

Vol. 38 No. 13: AAAI-24 Technical Tracks 13

Section

AAAI Technical Track on Machine Learning IV