Searcharxiv⌕ Search

arXiv subjects

Lingjuan Lyu

Publications and source records attributed to Lingjuan Lyu.

At least 145 records · Page 8Linked to original sources
arXiv2022

Threats to Pre-trained Language Models: Survey and Taxonomy

Pre-trained language models (PTLMs) have achieved great success and remarkable performance over a wide range of natural language processing (NLP) tasks. However, there are also growing concerns regarding the potential security issues in the adoption of PTLMs. In this survey, we comprehensively systematize recently discovered threats to PTLM systems and applications. We perform our attack characterization from three interesting perspectives. (1) We show threats can occur at different stages of the PTLM pipeline raised by different malicious entities. (2) We identify two types of model transferability (landscape, portrait) that facilitate attacks. (3) Based on the attack goals, we summarize four categories of attacks (backdoor, evasion, data privacy and model privacy). We also discuss some open problems and research directions. We believe our survey and taxonomy will inspire future studies towards secure and privacy-preserving PTLMs.

cs.CR↗
arXiv2022

Differential Private Knowledge Transfer for Privacy-Preserving Cross-Domain Recommendation

Cross Domain Recommendation (CDR) has been popularly studied to alleviate the cold-start and data sparsity problem commonly existed in recommender systems. CDR models can improve the recommendation performance of a target domain by leveraging the data of other source domains. However, most existing CDR models assume information can directly 'transfer across the bridge', ignoring the privacy issues. To solve the privacy concern in CDR, in this paper, we propose a novel two stage based privacy-preserving CDR framework (PriCDR). In the first stage, we propose two methods, i.e., Johnson-Lindenstrauss Transform (JLT) based and Sparse-awareJLT (SJLT) based, to publish the rating matrix of the source domain using differential privacy. We theoretically analyze the privacy and utility of our proposed differential privacy based rating publishing methods. In the second stage, we propose a novel heterogeneous CDR model (HeteroCDR), which uses deep auto-encoder and deep neural network to model the published source rating matrix and target rating matrix respectively. To this end, PriCDR can not only protect the data privacy of the source domain, but also alleviate the data sparsity of the source domain. We conduct experiments on two benchmark datasets and the results demonstrate the effectiveness of our proposed PriCDR and HeteroCDR.

cs.LG↗
arXiv2022

Privacy and Robustness in Federated Learning: Attacks and Defenses

As data are increasingly being stored in different silos and societies becoming more aware of data privacy issues, the traditional centralized training of artificial intelligence (AI) models is facing efficiency and privacy challenges. Recently, federated learning (FL) has emerged as an alternative solution and continue to thrive in this new reality. Existing FL protocol design has been shown to be vulnerable to adversaries within or outside of the system, compromising data privacy and system robustness. Besides training powerful global models, it is of paramount importance to design FL systems that have privacy guarantees and are resistant to different types of adversaries. In this paper, we conduct the first comprehensive survey on this topic. Through a concise introduction to the concept of FL, and a unique taxonomy covering: 1) threat models; 2) poisoning attacks and defenses against robustness; 3) inference attacks and defenses against privacy, we provide an accessible review of this important topic. We highlight the intuitions, key techniques as well as fundamental assumptions adopted by various attacks and defenses. Finally, we discuss promising future research directions towards robust and privacy-preserving federated learning.

cs.CR↗
arXiv2021

Killing One Bird with Two Stones: Model Extraction and Attribute Inference Attacks against BERT-based APIs

The collection and availability of big data, combined with advances in pre-trained models (e.g., BERT, XLNET, etc), have revolutionized the predictive performance of modern natural language processing tasks, ranging from text classification to text generation. This allows corporations to provide machine learning as a service (MLaaS) by encapsulating fine-tuned BERT-based models as APIs. However, BERT-based APIs have exhibited a series of security and privacy vulnerabilities. For example, prior work has exploited the security issues of the BERT-based APIs through the adversarial examples crafted by the extracted model. However, the privacy leakage problems of the BERT-based APIs through the extracted model have not been well studied. On the other hand, due to the high capacity of BERT-based APIs, the fine-tuned model is easy to be overlearned, but what kind of information can be leaked from the extracted model remains unknown. In this work, we bridge this gap by first presenting an effective model extraction attack, where the adversary can practically steal a BERT-based API (the target/victim model) by only querying a limited number of queries. We further develop an effective attribute inference attack which can infer the sensitive attribute of the training data used by the BERT-based APIs. Our extensive experiments on benchmark datasets under various realistic settings validate the potential vulnerabilities of BERT-based APIs. Moreover, we demonstrate that two promising defense methods become ineffective against our attacks, which calls for more effective defense methods.

cs.CR↗
arXiv2021

Protecting Intellectual Property of Language Generation APIs with Lexical Watermark

Nowadays, due to the breakthrough in natural language generation (NLG), including machine translation, document summarization, image captioning, etc NLG models have been encapsulated in cloud APIs to serve over half a billion people worldwide and process over one hundred billion word generations per day. Thus, NLG APIs have already become essential profitable services in many commercial companies. Due to the substantial financial and intellectual investments, service providers adopt a pay-as-you-use policy to promote sustainable market growth. However, recent works have shown that cloud platforms suffer from financial losses imposed by model extraction attacks, which aim to imitate the functionality and utility of the victim services, thus violating the intellectual property (IP) of cloud APIs. This work targets at protecting IP of NLG APIs by identifying the attackers who have utilized watermarked responses from the victim NLG APIs. However, most existing watermarking techniques are not directly amenable for IP protection of NLG APIs. To bridge this gap, we first present a novel watermarking method for text generation APIs by conducting lexical modification to the original outputs. Compared with the competitive baselines, our watermark approach achieves better identifiable performance in terms of p-value, with fewer semantic losses. In addition, our watermarks are more understandable and intuitive to humans than the baselines. Finally, the empirical studies show our approach is also applicable to queries from different domains, and is effective on the attacker trained on a mixture of the corpus which includes less than 10\% watermarked samples.

cs.CR↗
arXiv2021

Anti-Backdoor Learning: Training Clean Models on Poisoned Data

Backdoor attack has emerged as a major security threat to deep neural networks (DNNs). While existing defense methods have demonstrated promising results on detecting or erasing backdoors, it is still not clear whether robust training methods can be devised to prevent the backdoor triggers being injected into the trained model in the first place. In this paper, we introduce the concept of \emph{anti-backdoor learning}, aiming to train \emph{clean} models given backdoor-poisoned data. We frame the overall learning process as a dual-task of learning the \emph{clean} and the \emph{backdoor} portions of data. From this view, we identify two inherent characteristics of backdoor attacks as their weaknesses: 1) the models learn backdoored data much faster than learning with clean data, and the stronger the attack the faster the model converges on backdoored data; 2) the backdoor task is tied to a specific class (the backdoor target class). Based on these two weaknesses, we propose a general learning scheme, Anti-Backdoor Learning (ABL), to automatically prevent backdoor attacks during training. ABL introduces a two-stage \emph{gradient ascent} mechanism for standard training to 1) help isolate backdoor examples at an early training stage, and 2) break the correlation between backdoor examples and the target class at a later training stage. Through extensive experiments on multiple benchmark datasets against 10 state-of-the-art attacks, we empirically show that ABL-trained models on backdoor-poisoned data achieve the same performance as they were trained on purely clean data. Code is available at \url{https://github.com/bboylyg/ABL}.

cs.LG↗
arXiv2021

A Novel Attribute Reconstruction Attack in Federated Learning

Federated learning (FL) emerged as a promising learning paradigm to enable a multitude of participants to construct a joint ML model without exposing their private training data. Existing FL designs have been shown to exhibit vulnerabilities which can be exploited by adversaries both within and outside of the system to compromise data privacy. However, most current works conduct attacks by leveraging gradients on a small batch of data, which is less practical in FL. In this work, we consider a more practical and interesting scenario in which participants share their epoch-averaged gradients (share gradients after at least 1 epoch of local training) rather than per-example or small batch-averaged gradients as in previous works. We perform the first systematic evaluation of attribute reconstruction attack (ARA) launched by the malicious server in the FL system, and empirically demonstrate that the shared epoch-averaged local model gradients can reveal sensitive attributes of local training data of any victim participant. To achieve this goal, we develop a more effective and efficient gradient matching based method called cos-matching to reconstruct the training data attributes. We evaluate our attacks on a variety of real-world datasets, scenarios, assumptions. Our experiments show that our proposed method achieves better attribute attack performance than most existing baselines.

cs.CR↗
arXiv2021

A Reputation Mechanism Is All You Need: Collaborative Fairness and Adversarial Robustness in Federated Learning

Federated learning (FL) is an emerging practical framework for effective and scalable machine learning among multiple participants, such as end users, organizations and companies. However, most existing FL or distributed learning frameworks have not well addressed two important issues together: collaborative fairness and adversarial robustness (e.g. free-riders and malicious participants). In conventional FL, all participants receive the global model (equal rewards), which might be unfair to the high-contributing participants. Furthermore, due to the lack of a safeguard mechanism, free-riders or malicious adversaries could game the system to access the global model for free or to sabotage it. In this paper, we propose a novel Robust and Fair Federated Learning (RFFL) framework to achieve collaborative fairness and adversarial robustness simultaneously via a reputation mechanism. RFFL maintains a reputation for each participant by examining their contributions via their uploaded gradients (using vector similarity) and thus identifies non-contributing or malicious participants to be removed. Our approach differentiates itself by not requiring any auxiliary/validation dataset. Extensive experiments on benchmark datasets show that RFFL can achieve high fairness and is very robust to different types of adversaries while achieving competitive predictive accuracy.

cs.LG↗
arXiv2021

A Vertical Federated Learning Framework for Graph Convolutional Network

Recently, Graph Neural Network (GNN) has achieved remarkable success in various real-world problems on graph data. However in most industries, data exists in the form of isolated islands and the data privacy and security is also an important issue. In this paper, we propose FedVGCN, a federated GCN learning paradigm for privacy-preserving node classification task under data vertically partitioned setting, which can be generalized to existing GCN models. Specifically, we split the computation graph data into two parts. For each iteration of the training process, the two parties transfer intermediate results to each other under homomorphic encryption. We conduct experiments on benchmark data and the results demonstrate the effectiveness of FedVGCN in the case of GraphSage.

cs.LG↗
arXiv2021

Federated Model Distillation with Noise-Free Differential Privacy

Conventional federated learning directly averages model weights, which is only possible for collaboration between models with homogeneous architectures. Sharing prediction instead of weight removes this obstacle and eliminates the risk of white-box inference attacks in conventional federated learning. However, the predictions from local models are sensitive and would leak training data privacy to the public. To address this issue, one naive approach is adding the differentially private random noise to the predictions, which however brings a substantial trade-off between privacy budget and model performance. In this paper, we propose a novel framework called FEDMD-NFDP, which applies a Noise-Free Differential Privacy (NFDP) mechanism into a federated model distillation framework. Our extensive experimental results on various datasets validate that FEDMD-NFDP can deliver not only comparable utility and communication efficiency but also provide a noise-free differential privacy guarantee. We also demonstrate the feasibility of our FEDMD-NFDP by considering both IID and non-IID setting, heterogeneous model architectures, and unlabelled public datasets from a different distribution.

cs.CR↗
arXiv2021

DP-SIGNSGD: When Efficiency Meets Privacy and Robustness

Federated learning (FL) has emerged as a promising collaboration paradigm by enabling a multitude of parties to construct a joint model without exposing their private training data. Three main challenges in FL are efficiency, privacy, and robustness. The recently proposed SIGNSGD with majority vote shows a promising direction to deal with efficiency and Byzantine robustness. However, there is no guarantee that SIGNSGD is privacy-preserving. In this paper, we bridge this gap by presenting an improved method called DP-SIGNSGD, which can meet all the aforementioned properties. We further propose an error-feedback variant of DP-SIGNSGD to improve accuracy. Experimental results on benchmark image datasets demonstrate the effectiveness of our proposed methods.

cs.CR↗
arXiv2021

Robust Training Using Natural Transformation

Previous robustness approaches for deep learning models such as data augmentation techniques via data transformation or adversarial training cannot capture real-world variations that preserve the semantics of the input, such as a change in lighting conditions. To bridge this gap, we present NaTra, an adversarial training scheme that is designed to improve the robustness of image classification algorithms. We target attributes of the input images that are independent of the class identification, and manipulate those attributes to mimic real-world natural transformations (NaTra) of the inputs, which are then used to augment the training dataset of the image classifier. Specifically, we apply \textit{Batch Inverse Encoding and Shifting} to map a batch of given images to corresponding disentangled latent codes of well-trained generative models. \textit{Latent Codes Expansion} is used to boost image reconstruction quality through the incorporation of extended feature maps. \textit{Unsupervised Attribute Directing and Manipulation} enables identification of the latent directions that correspond to specific attribute changes, and then produce interpretable manipulations of those attributes, thereby generating natural transformations to the input data. We demonstrate the efficacy of our scheme by utilizing the disentangled latent representations derived from well-trained GANs to mimic transformations of an image that are similar to real-world natural variations (such as lighting conditions or hairstyle), and train models to be invariant to these natural transformations. Extensive experiments show that our method improves generalization of classification models and increases its robustness to various real-world distortions

cs.CV↗
arXiv2021

Model Extraction and Adversarial Transferability, Your BERT is Vulnerable!

Natural language processing (NLP) tasks, ranging from text classification to text generation, have been revolutionised by the pre-trained language models, such as BERT. This allows corporations to easily build powerful APIs by encapsulating fine-tuned BERT models for downstream tasks. However, when a fine-tuned BERT model is deployed as a service, it may suffer from different attacks launched by malicious users. In this work, we first present how an adversary can steal a BERT-based API service (the victim/target model) on multiple benchmark datasets with limited prior knowledge and queries. We further show that the extracted model can lead to highly transferable adversarial attacks against the victim model. Our studies indicate that the potential vulnerabilities of BERT-based API services still hold, even when there is an architectural mismatch between the victim model and the attack model. Finally, we investigate two defence strategies to protect the victim model and find that unless the performance of the victim model is sacrificed, both model ex-traction and adversarial transferability can effectively compromise the target models

cs.CL↗
arXiv2021

Privacy-Preserving Blockchain-Based Federated Learning for IoT Devices

Home appliance manufacturers strive to obtain feedback from users to improve their products and services to build a smart home system. To help manufacturers develop a smart home system, we design a federated learning (FL) system leveraging the reputation mechanism to assist home appliance manufacturers to train a machine learning model based on customers' data. Then, manufacturers can predict customers' requirements and consumption behaviors in the future. The working flow of the system includes two stages: in the first stage, customers train the initial model provided by the manufacturer using both the mobile phone and the mobile edge computing (MEC) server. Customers collect data from various home appliances using phones, and then they download and train the initial model with their local data. After deriving local models, customers sign on their models and send them to the blockchain. In case customers or manufacturers are malicious, we use the blockchain to replace the centralized aggregator in the traditional FL system. Since records on the blockchain are untampered, malicious customers or manufacturers' activities are traceable. In the second stage, manufacturers select customers or organizations as miners for calculating the averaged model using received models from customers. By the end of the crowdsourcing task, one of the miners, who is selected as the temporary leader, uploads the model to the blockchain. To protect customers' privacy and improve the test accuracy, we enforce differential privacy on the extracted features and propose a new normalization technique. We experimentally demonstrate that our normalization technique outperforms batch normalization when features are under differential privacy protection. In addition, to attract more customers to participate in the crowdsourcing FL task, we design an incentive mechanism to award participants.

cs.CR↗
arXiv2021

Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks

Deep neural networks (DNNs) are known vulnerable to backdoor attacks, a training time attack that injects a trigger pattern into a small proportion of training data so as to control the model's prediction at the test time. Backdoor attacks are notably dangerous since they do not affect the model's performance on clean examples, yet can fool the model to make incorrect prediction whenever the trigger pattern appears during testing. In this paper, we propose a novel defense framework Neural Attention Distillation (NAD) to erase backdoor triggers from backdoored DNNs. NAD utilizes a teacher network to guide the finetuning of the backdoored student network on a small clean subset of data such that the intermediate-layer attention of the student network aligns with that of the teacher network. The teacher network can be obtained by an independent finetuning process on the same clean subset. We empirically show, against 6 state-of-the-art backdoor attacks, NAD can effectively erase the backdoor triggers using only 5\% clean training data without causing obvious performance degradation on clean examples. Code is available in https://github.com/bboylyg/NAD.

cs.LG↗
arXiv2020

Local Differential Privacy based Federated Learning for Internet of Things

Internet of Vehicles (IoV) is a promising branch of the Internet of Things. IoV simulates a large variety of crowdsourcing applications such as Waze, Uber, and Amazon Mechanical Turk, etc. Users of these applications report the real-time traffic information to the cloud server which trains a machine learning model based on traffic information reported by users for intelligent traffic management. However, crowdsourcing application owners can easily infer users' location information, which raises severe location privacy concerns of the users. In addition, as the number of vehicles increases, the frequent communication between vehicles and the cloud server incurs unexpected amount of communication cost. To avoid the privacy threat and reduce the communication cost, in this paper, we propose to integrate federated learning and local differential privacy (LDP) to facilitate the crowdsourcing applications to achieve the machine learning model. Specifically, we propose four LDP mechanisms to perturb gradients generated by vehicles. The Three-Outputs mechanism is proposed which introduces three different output possibilities to deliver a high accuracy when the privacy budget is small. The output possibilities of Three-Outputs can be encoded with two bits to reduce the communication cost. Besides, to maximize the performance when the privacy budget is large, an optimal piecewise mechanism (PM-OPT) is proposed. We further propose a suboptimal mechanism (PM-SUB) with a simple formula and comparable utility to PM-OPT. Then, we build a novel hybrid mechanism by combining Three-Outputs and PM-SUB.

cs.CR↗
arXiv2020

A Fast and Scalable Authentication Scheme in IoT for Smart Living

Numerous resource-limited smart objects (SOs) such as sensors and actuators have been widely deployed in smart environments, opening new attack surfaces to intruders. The severe security flaw discourages the adoption of the Internet of things in smart living. In this paper, we leverage fog computing and microservice to push certificate authority (CA) functions to the proximity of data sources. Through which, we can minimize attack surfaces and authentication latency, and result in a fast and scalable scheme in authenticating a large volume of resource-limited devices. Then, we design lightweight protocols to implement the scheme, where both a high level of security and low computation workloads on SO (no bilinear pairing requirement on the client-side) is accomplished. Evaluations demonstrate the efficiency and effectiveness of our scheme in handling authentication and registration for a large number of nodes, meanwhile protecting them against various threats to smart living. Finally, we showcase the success of computing intelligence movement towards data sources in handling complicated services.

cs.CR↗
arXiv2020

Differentially Private Representation for NLP: Formal Guarantee and An Empirical Study on Privacy and Fairness

It has been demonstrated that hidden representation learned by a deep model can encode private information of the input, hence can be exploited to recover such information with reasonable accuracy. To address this issue, we propose a novel approach called Differentially Private Neural Representation (DPNR) to preserve the privacy of the extracted representation from text. DPNR utilises Differential Privacy (DP) to provide a formal privacy guarantee. Further, we show that masking words via dropout can further enhance privacy. To maintain utility of the learned representation, we integrate DP-noisy representation into a robust training process to derive a robust target model, which also helps for model fairness over various demographic variables. Experimental results on benchmark datasets under various parameter settings demonstrate that DPNR largely reduces privacy leakage without significantly sacrificing the main task performance.

cs.LG↗