CRMar 11, 2021

TAG: Gradient Attack on Transformer-based Language Models

arXiv:2103.06819v6669 citations
Originality Highly original
AI Analysis

This addresses privacy risks for users of Transformer-based NLP models in federated learning settings, representing a novel investigation into gradient attacks on language models.

The paper tackles the problem of gradient leakage attacks on Transformer-based language models in federated learning, proposing the TAG algorithm which achieves up to 90% data recovery on the CoLA dataset and outperforms prior methods with 1.5x recover rate and 2.5x ROUGE-2 scores.

Although federated learning has increasingly gained attention in terms of effectively utilizing local devices for data privacy enhancement, recent studies show that publicly shared gradients in the training process can reveal the private training images (gradient leakage) to a third-party in computer vision. We have, however, no systematic understanding of the gradient leakage mechanism on the Transformer based language models. In this paper, as the first attempt, we formulate the gradient attack problem on the Transformer-based language models and propose a gradient attack algorithm, TAG, to reconstruct the local training data. We develop a set of metrics to evaluate the effectiveness of the proposed attack algorithm quantitatively. Experimental results on Transformer, TinyBERT$_{4}$, TinyBERT$_{6}$, BERT$_{BASE}$, and BERT$_{LARGE}$ using GLUE benchmark show that TAG works well on more weight distributions in reconstructing training data and achieves 1.5$\times$ recover rate and 2.5$\times$ ROUGE-2 over prior methods without the need of ground truth label. TAG can obtain up to 90$\%$ data by attacking gradients in CoLA dataset. In addition, TAG has a stronger adversary on large models, small dictionary size, and small input length. We hope the proposed TAG will shed some light on the privacy leakage problem in Transformer-based NLP models.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes