mlsanityData to 7 Oct 2026

Variational Continual Test-Time Adaptation

arXiv 2402.08182 ↗First posted 13 Feb 2024cs.LG stat.MLcited by 7

Not on the map. Why?

Abstract (arXiv)

Continual Test-Time Adaptation (CTTA) task investigates effective domain adaptation under the scenario of continuous domain shifts during testing time. Due to the utilization of solely unlabeled samples, there exists significant uncertainty in model updates, leading CTTA to encounter severe error accumulation issues. In this paper, we introduce VCoTTA, a variational Bayesian approach to measure uncertainties in CTTA. At the source stage, we transform a pretrained deterministic model into a Bayesian Neural Network (BNN) via a variational warm-up strategy, injecting uncertainties into the model. During the testing time, we employ a mean-teacher update strategy using variational inference for the student model and exponential moving average for the teacher model. Our novel approach updates the student model by combining priors from both the source and teacher models. The evidence lower bound is formulated as the cross-entropy between the student and teacher models, along with the Kullback-Leibler (KL) divergence of the prior mixture. Experimental results on three datasets demonstrate the method's effectiveness in mitigating error accumulation within the CTTA framework.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 43 of its 77 references that are in the map. First posted on arXiv 13 Feb 2024.

Builds on

43 of 77 references.

References · 43

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
  3. 2016Wide Residual Networkscited by 3,057
  4. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,540
  5. 2016Learning without Forgettingcited by 2,465
  6. 2019Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shiftcited by 1,274
  7. 2018Deep Clustering for Unsupervised Learning of Visual Featurescited by 1,020
  8. 2019A Simple Baseline for Bayesian Uncertainty in Deep Learningcited by 555
  9. 2022Continual Test-Time Domain Adaptationcited by 520
  10. 2019Symmetric Cross Entropy for Robust Learning with Noisy Labelscited by 499
  11. 2015Variational Dropout and the Local Reparameterization Trickcited by 489
  12. 2022Efficient Test-Time Model Adaptation without Forgettingcited by 467
31 more
  1. 2020Improving robustness against common corruptions by covariate shift adaptationcited by 427
  2. 2021MEMO: Test Time Robustness via Adaptation and Augmentationcited by 376
  3. 2017Variational Continual Learningcited by 335
  4. 2015Probabilistic Backpropagation for Scalable Learning of Bayesian Neural Networkscited by 295
  5. 2022Contrastive Test-Time Adaptationcited by 258
  6. 2022Test-Time Training with Masked Autoencoderscited by 208
  7. 2023Robust Test-Time Adaptation in Dynamic Scenarioscited by 185
  8. 2017Multiplicative Normalizing Flows for Variational Bayesian Neural Networkscited by 157
  9. 2023EcoTTA: Memory-Efficient Continual Test-time Adaptation via Self-distilled Regularizationcited by 123
  10. 2021Bayesian Neural Network Priors Revisitedcited by 121
  11. 2022Robust Mean Teacher for Continual and Gradual Test-Time Adaptationcited by 102
  12. 2019Uncertainty-guided Continual Learning with Bayesian Neural Networkscited by 100
  13. 2022Decorate the Newcomers: Visual Domain Prompt for Continual Test Time Adaptationcited by 98
  14. 2020Domain Adaptation without Source Datacited by 88
  15. 2021Test-Time Adaptation to Distribution Shift by Confidence Maximization and Input Transformationcited by 67
  16. 2022Improving Test-Time Adaptation via Shift-agnostic Weight Regularization and Nearest Source Prototypescited by 67
  17. 2023ViDA: Homeostatic Visual Domain Adapter for Continual Test Time Adaptationcited by 58
  18. 2022Exploring Example Influence in Continual Learningcited by 44
  19. 2022A Probabilistic Framework for Lifelong Test-Time Adaptationcited by 44
  20. 2016Deep Gaussian Processes for Regression using Approximate Expectation Propagationcited by 39
  21. 2019A Unifying Bayesian View of Continual Learningcited by 38
  22. 2023Continual Test-time Domain Adaptation via Dynamic Sample Selectioncited by 35
  23. 2023Continual-MAE: Adaptive Distribution Masked Autoencoders for Continual Test-Time Adaptationcited by 30
  24. 2023Each Test Image Deserves A Specific Prompt: Continual Test-Time Adaptation for 2D Medical Image Segmentationcited by 29
  25. 2020Multi-Domain Multi-Task Rehearsal for Lifelong Learningcited by 24
  26. 2021Deep Bayesian Unsupervised Lifelong Learningcited by 21
  27. 2022CAFA: Class-Aware Feature Alignment for Test-Time Adaptationcited by 21
  28. 2022Multi-Label Continual Learning using Augmented Graph Convolutional Networkcited by 12
  29. 2023SATA: Source Anchoring and Target Alignment Network for Continual Test Time Adaptationcited by 8
  30. 2023Centroid Distance Distillation for Effective Rehearsal in Continual Learningcited by 4
  31. 2023Test-Time Adaptation for Nighttime Color-Thermal Semantic Segmentationcited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 4; 2025: 2; 2026: 1.

7 papers.

  1. 2024Confidence Self-Calibration for Multi-Label Class-Incremental Learningcited by 6
  2. 2024Controllable Continual Test-Time Adaptationcited by 6
  3. 2024Parameter-Selective Continual Test-Time Adaptationcited by 6
  4. 2025APCoTTA: Continual Test-Time Adaptation for Semantic Segmentation of Airborne LiDAR Point Cloudscited by 2
  5. 2025Test-Time Discovery via Hashing Memorycited by 1
  6. 2024Towards stable training of parallel continual learningcited by 0
  7. 2026Sample-wise Targeted Adversarial Attacks on Test-time Adaptationcited by 0