mlsanityData to 14 Oct 2026

Plug-In Inversion: Model-Agnostic Inversion for Vision with Data Augmentations

arXiv 2201.12961 ↗First posted 31 Jan 2022cs.CV cs.LGcited by 18

Not on the map. Why?

Abstract (arXiv)

Existing techniques for model inversion typically rely on hard-to-tune regularizers, such as total variation or feature regularization, which must be individually calibrated for each network in order to produce adequate images. In this work, we introduce Plug-In Inversion, which relies on a simple set of augmentations and does not require excessive hyper-parameter tuning. Under our proposed augmentation-based scheme, the same set of augmentation hyper-parameters can be used for inverting a wide range of image classification models, regardless of input dimensions or the architecture. We illustrate the practicality of our approach by inverting Vision Transformers (ViTs) and Multi-Layer Perceptrons (MLPs) trained on the ImageNet dataset, tasks which to the best of our knowledge have not been successfully accomplished by any previous works.

Key facts. Cited by 18 papers in the map (data through 14 Oct 2026). Builds on 36 of its 49 references that are in the map. First posted on arXiv 31 Jan 2022.

Builds on

36 of 49 references.

References · 36

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  4. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  5. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  6. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  7. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
  8. 2016Densely Connected Convolutional Networkscited by 6,720
  9. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
  10. 2020Training data-efficient image transformers & distillation through attentioncited by 4,586
  11. 2014Going Deeper with Convolutionscited by 4,241
  12. 2016Wide Residual Networkscited by 3,070
24 more
  1. 2016Improved Techniques for Training GANscited by 2,970
  2. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,546
  3. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,371
  4. 2013Visualizing and Understanding Convolutional Networkscited by 2,290
  5. 2020Improved Baselines with Momentum Contrastive Learningcited by 2,287
  6. 2019Searching for MobileNetV3cited by 1,994
  7. 2021MLP-Mixer: An all-MLP Architecture for Visioncited by 1,446
  8. 2018ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Designcited by 1,136
  9. 2016SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model sizecited by 1,060
  10. 2018MnasNet: Platform-Aware Neural Architecture Search for Mobilecited by 943
  11. 2018Robustness May Be at Odds with Accuracycited by 880
  12. 2019Adversarial Training for Free!cited by 643
  13. 2021Twins: Revisiting the Design of Spatial Attention in Vision Transformerscited by 565
  14. 2021ConViT: Improving Vision Transformers with Soft Convolutional Inductive Biasescited by 484
  15. 2019Dreaming to Distill: Data-free Knowledge Transfer via DeepInversioncited by 423
  16. 2014Understanding Deep Image Representations by Inverting Themcited by 403
  17. 2021Pay Attention to MLPscited by 368
  18. 2021Rethinking Spatial Dimensions of Vision Transformerscited by 361
  19. 2021ResMLP: Feedforward networks for image classification with data-efficient trainingcited by 339
  20. 2021Co-Scale Conv-Attentional Image Transformerscited by 224
  21. 2015Inverting Visual Representations with Convolutional Networkscited by 170
  22. 2021Sleeper Agent: Scalable Hidden Trigger Backdoors for Neural Networks Trained from Scratchcited by 102
  23. 2019Are Perceptually-Aligned Gradients a General Property of Robust Classifiers?cited by 43
  24. 2019Robust or Private? Adversarial Training Makes Models More Vulnerable to Privacy Attackscited by 11

Built on by

Cited by year

Citing papers per year: 2022: 2; 2023: 4; 2024: 7; 2025: 5.

18 papers.

Sort:

By area:

  1. 2022Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Modelscited by 424
  2. 2022What do Vision Transformers Learn? A Visual Explorationcited by 76
  3. 2025Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignmentcited by 62
  4. 2023Unlocking Feature Visualization for Deeper Networks with MAgnitude Constrained Optimizationcited by 27
  5. 2023Deconstructing Data Reconstruction: Multiclass, Weight Decay and General Lossescited by 26
  6. 2024What do we learn from inverting CLIP models?cited by 13
  7. 2024Model Inversion Attacks: A Survey of Approaches and Countermeasurescited by 11
  8. 2024U Can't Gen This? A Survey of Intellectual Property Protection Methods for Data in Generative AIcited by 7
  9. 2024Feature Accentuation: Revealing 'What' Features Respond to in Natural Imagescited by 5
  10. 2025VITAL: More Understandable Feature Visualization through Distribution Alignment and Relevant Information Flowcited by 3
  11. 2023Leaping Into Memories: Space-Time Deep Feature Synthesiscited by 2
  12. 2024LazyDP: Co-Designing Algorithm-Software for Scalable Training of Differentially Private Recommendation Modelscited by 1
6 more
  1. 2025MIMIC: Multimodal Inversion for Model Interpretation and Conceptualizationcited by 1
  2. 2023NeuroInspect: Interpretable Neuron-based Debugging Framework through Class-conditional Visualizationscited by 0
  3. 2024On mitigating stability-plasticity dilemma in CLIP-guided image morphing via geodesic distillation losscited by 0
  4. 2024DataFreeShield: Defending Adversarial Attacks without Training Datacited by 0
  5. 2025When Model Knowledge meets Diffusion Model: Diffusion-assisted Data-free Image Synthesis with Alignment of Domain and Classcited by 0
  6. 2025TRANSPORTER: Transferring Visual Semantics from VLM Manifoldscited by 0