mlsanityData to 7 Oct 2026

Test-Time Training Can Close the Natural Distribution Shift Performance Gap in Deep Learning Based Compressed Sensing

arXiv 2204.07204 ↗First posted 14 Apr 2022eess.IVcited by 37

Not on the map. Why?

Abstract (arXiv)

Deep learning based image reconstruction methods outperform traditional methods. However, neural networks suffer from a performance drop when applied to images from a different distribution than the training images. For example, a model trained for reconstructing knees in accelerated magnetic resonance imaging (MRI) does not reconstruct brains well, even though the same network trained on brains reconstructs brains perfectly well. Thus there is a distribution shift performance gap for a given neural network, defined as the difference in performance when training on a distribution P and training on another distribution Q, and evaluating both models on Q. In this work, we propose a domain adaptation method for deep learning based compressive sensing that relies on self-supervision during training paired with test-time training at inference. We show that for four natural distribution shifts, this method essentially closes the distribution shift performance gap for state-of-the-art architectures for accelerated MRI.

Key facts. Cited by 37 papers in the map (data through 7 Oct 2026). First posted on arXiv 14 Apr 2022.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 3; 2023: 11; 2024: 9; 2025: 10; 2026: 4.

37 papers.

Sort:

By area:

  1. 2023A Comprehensive Survey on Test-Time Adaptation under Distribution Shiftscited by 353
  2. 2022Test-Time Adaptation via Conjugate Pseudo-labelscited by 98
  3. 2023Score-Based Generative Models for PET Image Reconstructioncited by 22
  4. 2025SLOT: Sample-specific Language Model Optimization at Test-timecited by 17
  5. 2023Steerable Conditional Diffusion for Out-of-Distribution Adaptation in Medical Image Reconstructioncited by 15
  6. 2023SMRD: SURE-based Robust MRI Reconstruction with Diffusion Modelscited by 15
  7. 2022Scaling Laws For Deep Learning Based Image Reconstructioncited by 14
  8. 2023Prior Mismatch and Adaptation in PnP-ADMM with a Nonconvex Convergence Analysiscited by 10
  9. 2023Robustness of Deep Learning for Accelerated MRI: Benefits of Diverse Training Datacited by 10
  10. 2022Deep Model-Based Architectures for Inverse Problems under Mismatched Priorscited by 8
  11. 2024Stability and Generalizability in SDE Diffusion Models with Measure-Preserving Dynamicscited by 5
  12. 2025Test-time Adaptation for Foundation Medical Segmentation Model without Parametric Updatescited by 5
25 more
  1. 2023Test-Time Training for Speechcited by 4
  2. 2023Joint Supervised and Self-supervised Learning for MRI Reconstructioncited by 4
  3. 2025Benchmarking Self-Supervised Learning Methods for Accelerated MRI Reconstructioncited by 4
  4. 2025Unsupervised Detection of Distribution Shift in Inverse Problems using Diffusion Modelscited by 4
  5. 2023Towards Architecture-Agnostic Untrained Network Priors for Image Reconstruction with Frequency Regularizationcited by 3
  6. 2024Overcoming Distribution Shifts in Plug-and-Play Methods with Test-Time Trainingcited by 3
  7. 2024Generalized Semi-Supervised Learning via Self-Supervised Feature Adaptationcited by 3
  8. 2024Test-time adaptation for image compression with distribution regularizationcited by 3
  9. 2024Pruning Unrolled Networks (PUN) at Initialization for MRI Reconstruction Improves Generalizationcited by 3
  10. 2025UGoDIT: Unsupervised Group Deep Image Prior Via Transferable Weightscited by 3
  11. 2023Bridging Generalization Gaps in High Content Imaging Through Online Self-Supervised Domain Adaptationcited by 2
  12. 2024MotionTTT: 2D Test-Time-Training Motion Estimation for 3D Motion Corrected MRIcited by 2
  13. 2023Image Reconstruction via Deep Image Prior Subspacescited by 1
  14. 2024Robustness and Exploration of Variational and Machine Learning Approaches to Inverse Problems: An Overviewcited by 1
  15. 2024Resolution-Robust 3D MRI Reconstruction with 2D Diffusion Priors: Diverse-Resolution Training Outperforms Interpolationcited by 1
  16. 2025Efficient Test-Time Adaptation through Latent Subspace Coefficients Searchcited by 1
  17. 2024Equivariant Test-Time Training with Operator Sketching for Imaging Inverse Problemscited by 0
  18. 2025TAMER: A Test-Time Adaptive MoE-Driven Framework for EHR Representation Learningcited by 0
  19. 2025Fast Equivariant Imaging: Accelerating Unsupervised Learning and Model Adaptation via Inexact Splittingcited by 0
  20. 2025SlowFast-SCI: Slow-Fast Deep Unfolding Learning for Spectral Compressive Imagingcited by 0
  21. 2025Uncover and Unlearn Nuisances: Agnostic Fully Test-Time Adaptationcited by 0
  22. 2026Perspective-Equivariant Fine-tuning for Multispectral Demosaicing without Ground Truthcited by 0
  23. 2026Accelerating Stroke MRI with Diffusion Probabilistic Models through Large-Scale Pre-training and Target-Specific Fine-Tuningcited by 0
  24. 2026Entropy Minimization without Model Collapse: Mitigating Prediction Bias in Medical Imagingcited by 0
  25. 2026Awaken, Then Scale: Tiny Adaptation for MRI Reconstructioncited by 0