mlsanityData to 7 Oct 2026

Rethinking Model Inversion Attacks With Patch-Wise Reconstruction

arXiv 2312.07040 ↗First posted 12 Dec 2023cs.AI cs.CRcited by 0

Not on the map. Why?

Abstract (arXiv)

Model inversion (MI) attacks aim to infer or reconstruct the training dataset through reverse-engineering from the target model's weights. Recently, significant advancements in generative models have enabled MI attacks to overcome challenges in producing photo-realistic replicas of the training dataset, a technique known as generative MI. The generative MI primarily focuses on identifying latent vectors that correspond to specific target labels, leveraging a generative model trained with an auxiliary dataset. However, an important aspect is often overlooked: the MI attacks fail if the pre-trained generative model lacks the coverage to create an image corresponding to the target label, especially when there is a significant difference between the target and auxiliary datasets. To address this gap, we propose the Patch-MI method, inspired by a jigsaw puzzle, which offers a novel probabilistic interpretation of MI attacks. Even with a dissimilar auxiliary dataset, our method effectively creates images that closely mimic the distribution of image patches in the target dataset by patch-based reconstruction. Moreover, we numerically demonstrate that the Patch-MI improves Top 1 attack accuracy by 5%p compared to existing methods.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 39 of its 54 references that are in the map. First posted on arXiv 12 Dec 2023.

Builds on

39 of 54 references.

References · 39

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  3. 2022Generative Adversarial Networkscited by 5,603
  4. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,773
  5. 2014Going Deeper with Convolutionscited by 4,231
  6. 2016Deep Learning with Differential Privacycited by 2,682
  7. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,540
  8. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,221
  9. 2015Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networkscited by 2,048
  10. 2016Least Squares Generative Adversarial Networkscited by 1,000
  11. 2019An Introduction to Variational Autoencoderscited by 997
  12. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 925
27 more
  1. 2016InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Netscited by 858
  2. 2020What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimationcited by 492
  3. 2020Reliable Fidelity and Diversity Metrics for Generative Modelscited by 425
  4. 2019Dreaming to Distill: Data-free Knowledge Transfer via DeepInversioncited by 419
  5. 2020Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networkscited by 341
  6. 2019The Secret Revealer: Generative Model-Inversion Attacks Against Deep Neural Networkscited by 241
  7. 2017EMNIST: an extension of MNIST to handwritten letterscited by 212
  8. 2017Machine Learning Models that Remember Too Muchcited by 165
  9. 2022Reconstructing Training Data from Trained Neural Networkscited by 129
  10. 2021ViTGAN: Training GANs with Vision Transformerscited by 110
  11. 2018Model Reconstruction from Model Explanationscited by 84
  12. 2015How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?cited by 76
  13. 2022Variational Model Inversion Attackscited by 75
  14. 2020Knowledge-Enriched Distributional Model Inversion Attackscited by 65
  15. 2022Plug & Play Attacks: Towards Robust and Flexible Model Inversion Attackscited by 58
  16. 2019Generalized Variational Inference: Three arguments for deriving new Posteriorscited by 57
  17. 2023Re-thinking Model Inversion Attacks Against Deep Neural Networkscited by 39
  18. 2023Reinforcement Learning-Based Black-Box Model Inversion Attackscited by 33
  19. 2019Adversarial Neural Network Inversion via Auxiliary Knowledge Alignmentcited by 30
  20. 2023Pseudo Label-Guided Model Inversion Attack via Conditional Generative Adversarial Networkcited by 28
  21. 2023Unstoppable Attack: Label-Only Model Inversion via Conditional Diffusion Modelcited by 14
  22. 2021Improving Differentially Private SGD via Randomly Sparsified Gradientscited by 8
  23. 2023Breaching FedMD: Image Recovery via Paired-Logits Inversion Attackcited by 7
  24. 2023Boosting Model Inversion Attacks with Adversarial Examplescited by 7
  25. 2021Reconstructing Training Data from Diverse ML Models by Ensemble Inversioncited by 5
  26. 2022GAN-based Domain Inference Attackcited by 4
  27. 2023Generator Born from Classifiercited by 4

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Model Inversion Attacks Through Target-Specific Conditional Diffusion Models
  2. 2025Revisiting Model Inversion Evaluation: From Misleading Standards to Reliable Privacy Assessment
  3. 2025Do Vision-Language Models Leak What They Learn? Adaptive Token-Weighted Model Inversion Attacks
  4. 2023Targeted Image Reconstruction by Sampling Pre-trained Diffusion Model
  5. 2024How to Backdoor Consistency Models?
  6. 2023Black-box Membership Inference Attacks against Fine-tuned Diffusion Models
  7. 2026Public Diffusion Models, Private Images: Key-Controlled Inversion for Conditional Reconstruction
  8. 2024Dual-Model Defense: Safeguarding Diffusion Models from Membership Inference Attacks through Disjoint Data Splitting