mlsanityData to 14 Oct 2026

Visualizing Adapted Knowledge in Domain Transfer

arXiv 2104.10602 ↗First posted 20 Apr 2021cs.CV cs.AIcited by 32

Not on the map. Why?

Abstract (arXiv)

A source model trained on source data and a target model learned through unsupervised domain adaptation (UDA) usually encode different knowledge. To understand the adaptation process, we portray their knowledge difference with image translation. Specifically, we feed a translated image and its original version to the two models respectively, formulating two branches. Through updating the translated image, we force similar outputs from the two branches. When such requirements are met, differences between the two images can compensate for and hence represent the knowledge difference between models. To enforce similar outputs from the two branches and depict the adapted knowledge, we propose a source-free image translation method that generates source-style images using only target images and the two models. We visualize the adapted knowledge on several datasets with different UDA methods and find that generated images successfully capture the style difference between the two domains. For application, we show that generated images enable further tuning of the target model without accessing source data. Code available at https://github.com/hou-yz/DA_visualization.

Key facts. Cited by 32 papers in the map (data through 14 Oct 2026). Builds on 37 of its 49 references that are in the map. First posted on arXiv 20 Apr 2021.

Builds on

37 of 49 references.

References · 37

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  4. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  5. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 6,013
  6. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  7. 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,527
  8. 2015Domain-Adversarial Training of Neural Networkscited by 3,044
  9. 2016Perceptual Losses for Real-Time Style Transfer and Super-Resolutioncited by 2,694
  10. 2017Arbitrary Style Transfer in Real-time with Adaptive Instance Normalizationcited by 2,010
  11. 2015Learning Transferable Features with Deep Adaptation Networkscited by 1,329
  12. 2017Adversarial Discriminative Domain Adaptationcited by 1,311
25 more
  1. 2017CyCADA: Cycle-Consistent Adversarial Domain Adaptationcited by 1,082
  2. 2019When Does Label Smoothing Help?cited by 988
  3. 2019Relational Knowledge Distillationcited by 889
  4. 2020Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptationcited by 848
  5. 2017Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networkscited by 707
  6. 2017Maximum Classifier Discrepancy for Unsupervised Domain Adaptationcited by 683
  7. 2016Deep Transfer Learning with Joint Adaptation Networkscited by 612
  8. 2019Similarity-Preserving Knowledge Distillationcited by 516
  9. 2017VisDA: The Visual Domain Adaptation Challengecited by 435
  10. 2019Dreaming to Distill: Data-free Knowledge Transfer via DeepInversioncited by 423
  11. 2016Domain Separation Networkscited by 320
  12. 2016Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networkscited by 308
  13. 2016Learning from Simulated and Unsupervised Images through Adversarial Trainingcited by 288
  14. 2020Universal Source-Free Domain Adaptationcited by 220
  15. 2016Coupled Generative Adversarial Networkscited by 216
  16. 2019Data-Free Learning of Student Networkscited by 204
  17. 2017Generate To Adapt: Aligning Domains using Generative Adversarial Networkscited by 183
  18. 2017Deep Photo Style Transfercited by 168
  19. 2017Demystifying Neural Style Transfercited by 152
  20. 2017Data-Free Knowledge Distillation for Deep Neural Networkscited by 136
  21. 2019Zero-shot Knowledge Transfer via Adversarial Belief Matchingcited by 129
  22. 2016Unsupervised Cross-Domain Image Generationcited by 125
  23. 2019Zero-Shot Knowledge Distillation in Deep Networkscited by 121
  24. 2019The Knowledge Within: Methods for Data-Free Model Compressioncited by 54
  25. 2020Source Free Domain Adaptation with Image Translationcited by 21

Built on by

Cited by year

Citing papers per year: 2021: 1; 2022: 12; 2023: 8; 2024: 4; 2025: 7.

32 papers.

Sort:

By area:

  1. 2023A Comprehensive Survey on Test-Time Adaptation under Distribution Shiftscited by 359
  2. 2022Source-Free Unsupervised Domain Adaptation: A Surveycited by 80
  3. 2022Confidence Score for Source-Free Unsupervised Domain Adaptationcited by 59
  4. 2021Source-Free Open Compound Domain Adaptation in Semantic Segmentationcited by 43
  5. 2022Cross-Domain Correlation Distillation for Unsupervised Domain Adaptation in Nighttime Semantic Segmentationcited by 40
  6. 2022Generative Domain Adaptation for Face Anti-Spoofingcited by 33
  7. 2023The Unreasonable Effectiveness of Large Language-Vision Models for Source-free Video Domain Adaptationcited by 16
  8. 2025Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understandingcited by 13
  9. 2022RAIN: RegulArization on Input and Network for Black-Box Domain Adaptationcited by 9
  10. 2022Learning Unbiased Transferability for Domain Adaptation by Uncertainty Modelingcited by 8
  11. 2025Personalized Feature Translation for Expression Recognition: An Efficient Source-Free Domain Adaptation Methodcited by 8
  12. 2022Commonality in Natural Images Rescues GANs: Pretraining GANs with Generic and Privacy-free Synthetic Datacited by 6
20 more
  1. 2022Cluster-level pseudo-labelling for source-free cross-domain facial expression recognitioncited by 5
  2. 2024Reliable Source Approximation: Source-Free Unsupervised Domain Adaptation for Vestibular Schwannoma MRI Segmentationcited by 5
  3. 2023Learning Transferable Conceptual Prototypes for Interpretable Unsupervised Domain Adaptationcited by 4
  4. 2024TiGDistill-BEV: Multi-view BEV 3D Object Detection via Target Inner-Geometry Learning Distillationcited by 4
  5. 2023AdaptGuard: Defending Against Universal Attacks for Model Adaptationcited by 3
  6. 2023Target-agnostic Source-free Domain Adaptation for Regression Taskscited by 3
  7. 2022Explaining Cross-Domain Recognition with Interpretable Deep Classifiercited by 2
  8. 2022Dual Moving Average Pseudo-Labeling for Source-Free Inductive Domain Adaptationcited by 2
  9. 2023Contrast and Clustering: Learning Neighborhood Pair Representation for Source-free Domain Adaptationcited by 2
  10. 2022Explainable Supervised Domain Adaptationcited by 1
  11. 2022Learning to Structure an Image with Few Colors and Beyondcited by 1
  12. 2023Rectifying Group Irregularities in Explanations for Distribution Shiftcited by 1
  13. 2023Trustworthy Representation Learning Across Domainscited by 1
  14. 2025An Investigation of Test-time Adaptation for Audio Classification under Background Noisecited by 1
  15. 2025PhysFire-WM: A Physics-Informed World Model for Emulating Fire Spread Dynamicscited by 1
  16. 2024High-order Neighborhoods Know More: HyperGraph Learning Meets Source-free Unsupervised Domain Adaptationcited by 0
  17. 2024A3: Active Adversarial Alignment for Source-Free Domain Adaptationcited by 0
  18. 2025AdaptiveCDM: Source-Free Few-Shot Domain Adaptation for Cell Detection in Microscopic Imagescited by 0
  19. 2025OT Score: An OT based Confidence Score for Prototype-Assisted Source Free Unsupervised Domain Adaptationcited by 0
  20. 2025Model Recycling Framework for Multi-Source Data-Free Supervised Transfer Learningcited by 0