mlsanityData to 14 Oct 2026

Deconstructing Classifiers: Towards A Data Reconstruction Attack Against Text Classification Models

arXiv 2306.13789 ↗First posted 23 Jun 2023cs.CL cs.CR cs.LGcited by 6

Not on the map. Why?

Abstract (arXiv)

Natural language processing (NLP) models have become increasingly popular in real-world applications, such as text classification. However, they are vulnerable to privacy attacks, including data reconstruction attacks that aim to extract the data used to train the model. Most previous studies on data reconstruction attacks have focused on LLM, while classification models were assumed to be more secure. In this work, we propose a new targeted data reconstruction attack called the Mix And Match attack, which takes advantage of the fact that most classification models are based on LLM. The Mix And Match attack uses the base model of the target model to generate candidate tokens and then prunes them using the classification head. We extensively demonstrate the effectiveness of the attack using both random and organic canaries. This work highlights the importance of considering the privacy risks associated with data reconstruction attacks in classification models and offers insights into possible leakages.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 23 of its 37 references that are in the map. First posted on arXiv 23 Jun 2023.

Builds on

23 of 37 references.

References · 23

  1. 2017Attention Is All You Needcited by 60,361
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
  4. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
  5. 2020Extracting Training Data from Large Language Modelscited by 2,189
  6. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 864
  7. 2022Quantifying Memorization Across Neural Language Modelscited by 763
  8. 2018ML-Leaks: Model and Data Independent Membership Inference Attacks and Defenses on Machine Learning Modelscited by 498
  9. 2019Does Learning Require Memorization? A Short Tale about a Long Tailcited by 460
  10. 2020Label-Only Membership Inference Attackscited by 388
  11. 2019White-box vs Black-box: Bayes Optimal Strategies for Membership Inferencecited by 271
  12. 2019Stolen Memories: Leveraging Model Memorization for Calibrated White-Box Membership Inferencecited by 175
11 more
  1. 2022Reconstructing Training Data with Informed Adversariescited by 148
  2. 2019Updates-Leak: Data Set Inference and Reconstruction Attacks in Online Learningcited by 122
  3. 2020When is Memorization of Irrelevant Training Data Necessary for High-Accuracy Learning?cited by 104
  4. 2021Does BERT Pretrained on Clinical Notes Reveal Sensitive Data?cited by 93
  5. 2018Understanding Membership Inferences on Well-Generalized Learning Modelscited by 90
  6. 2019Analyzing Information Leakage of Updates to Natural Language Modelscited by 90
  7. 2021Property Inference From Poisoningcited by 60
  8. 2020Leakage of Dataset Properties in Multi-Party Machine Learningcited by 47
  9. 2018Towards Demystifying Membership Inference Attackscited by 40
  10. 2021Privacy Regularization: Joint Privacy-Utility Optimization in Language Modelscited by 30
  11. 2022Privacy Leakage in Text Classification: A Data Extraction Approachcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 2; 2025: 3; 2026: 1.

6 papers.

  1. 2024Privacy Leakage on DNNs: A Survey of Model Inversion Attacks and Defensescited by 30
  2. 2024Privacy-preserving Fine-tuning of Large Language Models through Flatnesscited by 6
  3. 2025NLP Security and Ethics, in the Wildcited by 2
  4. 2025Multi-Agent Framework for Threat Mitigation and Resilience in AI-Based Systemscited by 2
  5. 2025Label Forensics: Interpreting Hard Labels in Black-Box Text Classifiercited by 0
  6. 2026Boundary-targeted Membership Inference Attacks on Safety Classifierscited by 0