REVIEW 14 cited by
Towards Stable Test-Time Adaptation in Dynamic Wild World
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Test-time adaptation (TTA) has shown to be effective at tackling distribution shifts between training and testing data by adapting a given model on test samples. However, the online model updating of TTA may be unstable and this is often a key obstacle preventing existing TTA methods from being deployed in the real world. Specifically, TTA may fail to improve or even harm the model performance when test data have: 1) mixed distribution shifts, 2) small batch sizes, and 3) online imbalanced label distribution shifts, which are quite common in practice. In this paper, we investigate the unstable reasons and find that the batch norm layer is a crucial factor hindering TTA stability. Conversely, TTA can perform more stably with batch-agnostic norm layers, \ie, group or layer norm. However, we observe that TTA with group and layer norms does not always succeed and still suffers many failure cases. By digging into the failure cases, we find that certain noisy test samples with large gradients may disturb the model adaption and result in collapsed trivial solutions, \ie, assigning the same class label for all samples. To address the above collapse issue, we propose a sharpness-aware and reliable entropy minimization method, called SAR, for further stabilizing TTA from two aspects: 1) remove partial noisy samples with large gradients, 2) encourage model weights to go to a flat minimum so that the model is robust to the remaining noisy samples. Promising results demonstrate that SAR performs more stably over prior methods and is computationally efficient under the above wild test scenarios.
Forward citations
Cited by 14 Pith papers
-
Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models
ZAEC anchors calibration to each sample's zero-shot entropy and selectively softens over-sharpened TTA predictions, reaching the lowest macro-average calibration error among evaluated post-hoc methods on ViT-B/16.
-
Test-Time Curriculum for Open-Set AIGC Detection
A curriculum-based test-time adaptation method, using balanced confident pseudo-labels and multi-scale refinement, improves AIGC detector accuracy on unseen generators by 11 to 29 points over its starting detector.
-
When Can Test-Time Adaptation Help Zero-Shot CT Vision-Language Models?
CARVE is a label-free, cardinality-aware test-time adaptation method that consistently improves multi-label CT diagnosis when the base model is already discriminative and input depth matches pretraining.
-
No Time Like the Present: Agentic Test-Time Training for LLM Agents
Continuous test-time training that downweights tokens in repeated n-grams improves multi-turn LLM agent success by up to about 5 points by limiting self-reinforcing drift.
-
When Small Guides Large: Cross-Model Co-Learning for Test-Time Adaptation
COCA shows that bidirectional cross-model co-learning during test-time adaptation lets even small models significantly improve large models' robustness to distribution shift.
-
Align the GAP: Prior-based Unified Multi-Task Remote Physiological Measurement Framework For Domain Generalization and Personalization
A prior-based unified framework, GAP, jointly handles multi-source domain generalization and per-user test-time adaptation for multi-task rPPG, beating prior domain generalization and test-time adaptation methods on s...
-
Frequency Composition for Compressed and Domain-Adaptive Neural Networks
Training quantized models on low-frequency images plus frequency-aware batch normalization at test time improves both compression and domain-shift robustness.
-
SloMo-Fast: Slow-Momentum and Fast-Adaptive Teachers for Source-Free Continual Test-Time Adaptation
SloMo-Fast, a source-free dual-teacher CTTA method with test-time prototypes, reports SOTA error rates across 11 settings and a new cyclic domain-repetition benchmark, but the appendix data contain serious inconsistencies.
-
Enhancing Novel View Synthesis from extremely sparse views with SfM-free 3D Gaussian Splatting Framework
SfM-free 3D Gaussian Splatting jointly estimates camera poses and geometry from two views and reports a 2.75 dB PSNR improvement over existing 3DGS methods on extremely sparse inputs.
-
Continual Adaptation: Environment-Conditional Parameter Generation for Object Detection in Dynamic Scenarios
A conditional diffusion model generates LoRA adapter parameters for object detectors at test time, improving continual domain adaptation accuracy by small margins over prior methods.
-
Single Image Test-Time Adaptation via Multi-View Co-Training
MuVi achieves near-supervised breast MRI tumor segmentation by adapting a pretrained 3D model on a single test image via multi-view co-training with entropy-thresholded pseudolabels.
-
PAID: Pairwise Angular-Invariant Decomposition for Continual Test-Time Adaptation
PAID proposes Householder-based orthogonal weight updates for continual test-time adaptation, claiming that preserving pairwise angular structure of pretrained weights is a useful prior, but the math and validation fo...
-
Billet Number Recognition Based on Test-Time Adaptation
A test-time adaptation plus prior-knowledge correction method raises billet number recognition accuracy from 58% to 80% on machine-printed and from 24% to 70% on handwritten samples in the authors' tests.
-
Generalizing vision-language models to novel domains: A comprehensive survey
A survey of VLM generalization literature organized by transferred module, with benchmark tables and a review of multimodal LLMs.
Discussion (0). Continue with ORCID to comment.