DP-BART for Privatized Text Rewriting under Local Differential Privacy
arXiv:2302.07636 · doi:10.18653/v1/2023.findings-acl.874
Abstract
Privatized text rewriting with local differential privacy (LDP) is a recent approach that enables sharing of sensitive textual documents while formally guaranteeing privacy protection to individuals. However, existing systems face several issues, such as formal mathematical flaws, unrealistic privacy guarantees, privatization of only individual words, as well as a lack of transparency and reproducibility. In this paper, we propose a new system 'DP-BART' that largely outperforms existing LDP systems. Our approach uses a novel clipping method, iterative pruning, and further training of internal representations which drastically reduces the amount of noise required for DP guarantees. We run experiments on five textual datasets of varying sizes, rewriting them at different privacy guarantees and evaluating the rewritten texts on downstream text classification tasks. Finally, we thoroughly discuss the privatized text rewriting approach and its limitations, including the problem of the strict text adjacency constraint in the LDP paradigm that leads to the high noise requirement.
arXiv commentAccepted at ACL Findings 2023
References in corpus (8)
- Auto-Encoding Variational Bayes
- Deep Learning with Differential Privacy
- On the difficulty of training Recurrent Neural Networks
- The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
- Differential Privacy: An Economic Method for Choosing Epsilon
- Differential Privacy in Natural Language Processing: The Story So Far
- How reparametrization trick broke differentially-private text representation learning
- DP-Rewrite: Towards Reproducibility and Transparency in Differentially Private Text Rewriting
Cited by in corpus (3)
- A Survey on Current Trends and Recent Advances in Text Anonymization
- The Double-edged Sword of LLM-based Data Reconstruction: Understanding and Mitigating Contextual Vulnerability in Word-level Differential Privacy Text Sanitization
- CTIGuardian: A Few-Shot Framework for Mitigating Privacy Leakage in Fine-Tuned LLMs