CAPE: Context-Aware Private Embeddings for Private Language Learning

Plant, Richard; Gkatzia, Dimitra; Giuffrida, Mario Valerio

doi:10.18653/v1/2021.emnlp-main.628

Cited by 6 publications

(10 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to remove the dependence of dimensionality in Unary Encoding mechanism, they propose an Optimized Multiple Encoding, which embeds vectors with a certain fixed size. Their post-processing procedure was then improved by (Plant et al, 2021). In (Plant et al, 2021) Habernal, 2021) also study privatizing word embeddings.…”

Section: Vanilla Dpmentioning

confidence: 99%

See 1 more Smart Citation

Differentially Private Natural Language Models: Recent Advances and Future Directions

Hu¹,

Habernal²,

Shen³

et al. 2023

Preprint

View full text Add to dashboard Cite

Recent developments in deep learning have led to great success in various natural language processing (NLP) tasks. However, these applications may involve data that contain sensitive information. Therefore, how to achieve good performance while also protect privacy of sensitive data is a crucial challenge in NLP. To preserve privacy, Differential Privacy (DP), which can prevent reconstruction attacks and protect against potential side knowledge, is becoming a de facto technique for private data analysis. In recent years, NLP in DP models (DP-NLP) has been studied from different perspectives, which deserves a comprehensive review. In this paper, we provide the first systematic review of recent advances on DP deep learning models in NLP. In particular, we first discuss some differences and additional challenges of DP-NLP compared with the standard DP deep learning. Then we investigate some existing work on DP-NLP and present its recent developments from two aspects: gradient perturbation based methods and embedding vector perturbation based methods. We also discuss some challenges and future directions of this topic.

show abstract

Section: Vanilla Dpmentioning

confidence: 99%

“…Their post-processing procedure was then improved by (Plant et al, 2021). In (Plant et al, 2021) Habernal, 2021) also study privatizing word embeddings. However, instead of using Unary Encoding or dropout, Krishna et al (2021) propose ADePT which is an auto-encoder based DP algorithm.…”

Section: Vanilla Dpmentioning

confidence: 99%

Differentially Private Natural Language Models: Recent Advances and Future Directions

Hu¹,

Habernal²,

Shen³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…To obscure the representation itself, we add random perturbations to each representation of a random input, as shown in Figure 2. Following Plant et al (2021), we adopt Laplace Noise as the random perturbation. Formally, the process of mixing encryption and representation encryption for x 1:k i is:…”

Section: Privacy Mixermentioning

confidence: 99%

“…One solution is to apply cryptographic techniques to the PLM Chen et al, 2022), but these methods often come with significant communi-cation costs and computational time, making it difficult to apply them in real-world scenarios. Another potential solution is to remove the private information in word representations (Pan et al, 2020) through adversarial training (Li et al, 2018;Coavoux et al, 2018;Plant et al, 2021) and differential privacy (Lyu et al, 2020a;Hoory et al, 2021;Yue et al, 2021). However, the private information in our scenario pertains to each word in the user's plaintext.…”

Section: Introductionmentioning

confidence: 99%

TextMixer: Mixing Multiple Inputs for Privacy-Preserving Inference

Zhou,

Lu,

et al. 2023

Findings of the Association for Computational Linguistics: EMNLP 2023

View full text Add to dashboard Cite

Pre-trained language models (PLMs) are often deployed as cloud services, enabling users to upload textual data and perform inference remotely. However, users' personal text often contains sensitive information, and sharing such data directly with the service providers can lead to serious privacy leakage. To address this problem, we introduce a novel privacy-preserving inference framework called TextMixer, which prevents plaintext leakage during the inference phase. Inspired by k-anonymity, TextMixer aims to obfuscate a user's private input by mixing it with multiple other inputs, thereby confounding potential privacy attackers. To achieve this, our approach involves: (1) proposing a novel encryption module, Privacy Mixer, which encrypts input from three distinct dimensions: mixing, representation, and position. ( 2) adopting a pre-trained Multi-input Multi-output network to handle mixed representations and obtain multiple predictions. (3) employing a Privacy Demixer to ensure only the user can decrypt the real output among the multiple predictions. Furthermore, we explore different ways to automatically generate synthetic inputs required for mixing. Experimental results on token and sentence classification tasks demonstrate that TextMixer greatly surpasses existing privacy-preserving methods in both performance and privacy.

show abstract

“…In the context of document AI, such privacy techniques may be applied under different settings. For instance, an organization providing document AI services may train the models under global privacy constraints [22,30] to safeguard its own private data, or under local privacy constraints [31,32], where each individual client only uploads a private augmented data to the service, leaving no fingerprint that can be traced to the client. On the other hand, federated learning [23,24] may be deployed for private aggregation of data across multiple organizations or clients.…”

Section: Introductionmentioning

confidence: 99%

Towards Privacy Preserved Document Image Classification - A Comprehensive Benchmark

Saifullah,

Mercier,

Agne

et al. 2023

Preprint

View full text Add to dashboard Cite

As data-driven AI systems become increasingly integrated into industry, concerns have recently arisen regarding potential privacy breaches and the inadvertent leakage of sensitive user data through the exploitation of these systems. In this paper, we explore the intersection of data privacy and AI-powered document analysis systems, presenting a comprehensive benchmark of well-known privacy-preserving methods for the task of document image classification. In particular, we investigate four different privacy methods---Differential Privacy (DP), Federated Learning (FL), Differentially Private Federated Learning (DP-FL), and Secure Multi-Party Computation (SMPC)---on two well-known document benchmark datasets, namely RVL-CDIP and Tobacco3482. Furthermore, we investigate the performance of each method under a variety of configurations for thorough benchmarking. Finally, the privacy strength of each approach is assessed by subjecting the private models to well-known membership inference attacks. Our results demonstrate that, with sufficient tuning of hyperparameters, Differential Privacy (DP) can achieve reasonable performance on the task of document image classification while also ensuring rigorous privacy constraints, both in standalone and federated learning setups. On the other hand, while FL-based approaches present less implementation complexity and incur little to no loss in performance on the task, they do not offer sufficient protection against privacy attacks. By rigorously benchmarking various privacy approaches, our study paves the way for integrating deep document classification models into industrial pipelines while meeting regulatory and ethical standards, including GDPR and the AI Act 2022.

show abstract

CAPE: Context-Aware Private Embeddings for Private Language Learning

Cited by 6 publications

References 32 publications

Differentially Private Natural Language Models: Recent Advances and Future Directions

Differentially Private Natural Language Models: Recent Advances and Future Directions

TextMixer: Mixing Multiple Inputs for Privacy-Preserving Inference

Towards Privacy Preserved Document Image Classification - A Comprehensive Benchmark

Contact Info

Product

Resources

About