Information Security Commons^™

The Privacy Paradox Of Llms: User Perceptions And The Reality Of Pii Leakage, Shuai Cheng, Haitao Xu, Shu Meng, Shuai Hao, Chuan Yue, Zhao Li

Computer Science Faculty Publications

Large language models (LLMs) are increasingly deployed, yet they introduce significant privacy risks by disclosing personally identifiable information (PII) during interactions. Although prior work has demonstrated the feasibility of extracting PII from LLMs, no comprehensive study has evaluated the actual extent of PII leakage across mainstream LLMs or investigated user perceptions, literacy, and behavioral responses to these risks. To address these gaps, we conduct a large-scale evaluation of PII leakage in popular LLMs, demonstrating that attackers can extract email addresses and phone numbers with high success rates. Through a mixed-methods study involving 20 interviews and 204 survey participants, we identify …

Exploring Large Language Models For Trustworthy Use: Insights From Research And Development, Sandeep Kalari, Sahithi Padidela, Vikas Ashok, Ravi Mukkamala

Computer Science Faculty Publications

Large Language Models (LLMs) are increasingly being adopted in a wide variety of domains, including sensitive domains such as healthcare and finance. However, persistent challenges such as unreliable data sources, privacy breaches, and hallucinated output continue to hinder their usage. We have experimented with several strategies to address these challenges. First, we developed BlockQwen, a blockchain-augmented framework that integrates decentralized trust validation, role-specific access control, and verifiable audit trails into the Qwen 2.5 LLM workflow. Second, we developed PrivAware, a multilayered privacy-enforcement framework, using a fine-tuned Flan-T5 model with self-attention masking, to safeguard data while maintaining high utility. Both systems …

Privacy-Preserving Federated Learning With Optimized Ensemble Weighting And Knowledge Distillation For Covid-19 Detection From Non-Iid Medical Imaging Data, Richard Annan, Hong Qin, Robert Newman, Madhuri Siddula, Letu Qingge

Computer Science Faculty Publications

Medical imaging enables rapid and accurate diagnosis of COVID-19, with CT scans proving especially effective. However, data privacy concerns limit collaborative model development across hospitals. To address this issue, we introduce a novel federated learning framework. It is referred to as Independent Knowledge Distillation with post-Ensemble Federated Learning (IKDEFL). Differential Privacy (DP) is integrated into the framework to improve privacy guarantees. Three DP mechanisms are evaluated. These include Fixed Gaussian, Gaussian Adaptive, and Tree Adaptive. The evaluation has been conducted on heterogeneous and Non-Independent and Identically Distributed (Non-IID) datasets. These datasets reflect real-world hospital scenarios. Results show that IKDEFL significantly …

An Investigation Of Federated Gnns Under Aggregation, Data Poisoning, And Differential Privacy For Icu Length-Of-Stay Prediction, Shakib Mahmud Dipto, Soumya Banerjee, Sandip Roy, Ahmad F. Al Musawi, Preetam Ghosh, Sachin Shetty, Pratip Rana

Computer Science Faculty Publications

Accurate prediction of ICU Length of Stay (LoS) is essential for clinical decision-making and healthcare resource management. Graph Neural Networks (GNNs), such as GraphSAGE, offer a natural fit by capturing patient data from Electronic Health Records (EHRs) through graph structures. However, the distributed and sensitive nature of this data raises both privacy and legal concerns regarding the aggregation and training of GNN models. This additionally leads to issues with data imbalance and model robustness. In this study, we perform an analysis of the Federated Graph Neural Network (GNN-FL) framework to enable decentralized learning on EHRs derived from the MIMIC-III dataset. …

Privshap: A Finer-Granularity Network Linearization Method For Private Inference, Xiangrui Xu, Zhenzhen Wang, Rui Ning, Chunsheng Xiu, Hongyi Wu

Computer Science Faculty Publications

Private inference applies cryptographic techniques like homomorphic encryption, garble circuit and secret sharing to keep both sides privacy in a client-server setting during inference. It is often hindered by the high communication overheads, especially at non-linear activation layers such as ReLU. Hence ReLU pruning has been widely recognized as an efficient way to accelerate private inference. Existing approaches to ReLU pruning typically rely on coarse hypothesis, which assume an inverse correlation between the importance of ReLU and linear layers or shallow activation layers have less importance for universal models, to assign the budgets according to the layer while preserving the …

Not Here, Go There: Analyzing Redirection Patterns On The Web, Kritika Garg, Sawood Alam, Dietrich Ayala, Michele C. Weigle, Michael L. Nelson

Computer Science Faculty Publications

URI redirections are integral to web management, supporting structural changes, SEO optimization, and security. However, their complexities affect usability, SEO performance, and digital preservation. This study analyzed 11 million unique redirecting URIs, following redirections up to 10 hops per URI, to uncover patterns and implications of redirection practices. Our findings revealed that 50% of the URIs terminated successfully, while 50% resulted in errors, including 0.06% exceeding 10 hops. Canonical redirects, such as HTTP to HTTPS transitions, were prevalent, reflecting adherence to SEO best practices. Non-canonical redirects, often involving domain or path changes, highlighted significant web migrations, rebranding, and security risks. …

Position: Benchmarking Is Broken - Don't Let Ai Be Its Own Judge, Zerui Cheng, Stella Wohnig, Ruchika Gupta, Samiul Alam, Tassallah Abdullahi, João Alves Ribeiro, Christian Nielsen-Garcia, Saif Mir, Siran Li, Jason Orender, Seyed Ali Bahrainian, Daniel Kirste, Aaron Gokaslan, Carsten Eickhoff, Pramod Viswanath, Ruben Wolff

Computer Science Faculty Publications

The meteoric rise of Artificial Intelligence (AI), with its rapidly expanding market capitalization, presents both transformative opportunities and critical challenges. Chief among these is the urgent need for a new, unified paradigm for trustworthy evaluation, as current benchmarks increasingly reveal critical vulnerabilities. Issues like data contamination and selective reporting by model developers fuel hype, while inadequate data quality control can lead to biased evaluations that, even if unintentionally, may favor specific approaches. As a flood of participants enters the AI space, this "Wild West" of assessment makes distinguishing genuine progress from exaggerated claims exceptionally difficult. Such ambiguity blurs scientific signals …

Sting: A Stealthy Backdoor Attack On Gnn-Based Malicious Domain Detection Via Dns Perturbations, Muhammad Anan, Mahmoud Nazzal, Abdallah Khreishah, Issa Khalil, Nhathai Phan, Ahmad Sawalmeh

Computer Science Faculty Publications

Detecting malicious Internet domains is essential for safeguarding against various online threats. The current approach to detecting malicious domains (MDD) employs a graph neural network (GNN) method, which uses DNS logs to construct heterogeneous graphs for determining the maliciousness of unknown domains. Despite its success, this method is vulnerable to data poisoning attacks where an adversary can manipulate specific graph nodes to implant a backdoor into the model during training. To showcase the vulnerability, we propose a stealthy trigger injection attack on node features and graph structure in MDD, dubbed (STING). The attacker carefully manipulates selected features and edges of …

Adversarially Attacking Graph Properties And Sparsification In Graph Learning, Chunjiang Zhu, Blake Gaines, Jing Deng, Jinbo Bi

Computer Science Faculty Publications

Graph neural networks and graph transformers explicitly or implicitly rely on fundamental properties of the underlying graph, such as spectral properties and shortest-path distances. However, it is still not clear how these graph properties are vulnerable to adversarial attacks and what impacts this has on the downstream graph learning. Moreover, while graph sparsification has been used to improve computational cost of learning over graphs, its susceptibility to adversarial attacks has not been studied. In this paper, we study adversarial attacks on graph properties and graph sparsification and their impacts on downstream graph learning, paving the way for how to protect …

Effective Pii Extraction From Llms Through Augmented Few-Shot Learning, Shuai Cheng, Shu Meng, Haitao Xu, Haoran Zhang, Shuai Hao, Chuan Yue, Wenrui Ma, Meng Han, Fang Zhang, Zhao Li

Computer Science Faculty Publications

Large Language Models (LLMs) exhibit strong natural language processing capabilities but also pose significant privacy risks, particularly regarding the leakage of Personally Identifiable Information (PII) embedded in their training data. Existing PII extraction methods suffer from the limitations of low success rates or impracticality for large-scale PII extraction. In this study, we propose a novel PII extraction approach based on enhanced few-shot learning techniques, which achieves efficient and cost-effective PII retrieval without relying on fine-tuning or jailbreaking. We evaluated our approach on both open-source and closed-source LLMs. The experimental results demonstrate that, for non-targeted PII extraction, the attack success rate …

Understanding Pii Leakage In Large Language Models: A Systematic Survey, Shuai Cheng, Zhao Li, Shu Meng, Mengxia Ren, Haitao Xu, Shuai Hao, Chuan Yue, Fang Zhang

Computer Science Faculty Publications

Large Language Models (LLMs) have demonstrated exceptional success across a variety of tasks, particularly in natural language processing, leading to their growing integration into numerous facets of daily life. However, this widespread deployment has raised substantial privacy concerns, especially regarding personally identifiable information (PII), which can be directly associated with specific individuals. The leakage of such information presents significant real-world privacy threats. In this paper, we conduct a systematic investigation into existing research on PII leakage in LLMs, encompassing commonly utilized PII datasets, evaluation metrics, and current studies on both PII leakage attacks and defensive strategies. Finally, we identify unresolved …

Mosaic: A Prune-And-Assemble Approach For Efficient Model Pruning In Privacy-Preserving Deep Learning, Yifei Cai, Qiao Zhang, Rui Ning, Chunsheng Xin, Hongyi Wu

Computer Science Faculty Publications

To enable common users to capitalize on the power of deep learning, Machine Learning as a Service (MLaaS) has been proposed in the literature, which opens powerful deep learning models of service providers to the public. To protect the data privacy of end users, as well as the model privacy of the server, several state-of-the-art privacy-preserving MLaaS frameworks have also been proposed. Nevertheless, despite the exquisite design of these frameworks to enhance computation efficiency, the computational cost remains expensive for practical applications. To improve the computation efficiency of deep learning (DL) models, model pruning has been adopted as a strategic …

Dial "N" For Nxdomain: The Scale, Origin, And Security Implications Of Dns Queries To Non-Existent Domains, Gunnan Liu, Lin Jin, Shuai Hao, Yubao Zhang, Daiping Liu, Angelos Stavrou, Haining Wang

Computer Science Faculty Publications

Non-Existent Domain (NXDomain) is one type of the Domain Name System (DNS) error responses, indicating that the queried domain name does not exist and cannot be resolved. Unfortunately, little research has focused on understanding why and how NXDomain responses are generated, utilized, and exploited. In this paper, we conduct the first comprehensive and systematic study on NXDomain by investigating its scale, origin, and security implications. Utilizing a large-scale passive DNS database, we identify 146,363,745,785 NXDomains queried by DNS users between 2014 and 2022. Within these 146 billion NXDomains, 91 million of them hold historic WHOIS records, of which 5.3 million …

Camouflaged Poisoning Attack On Graph Neural Networks, Chao Jiang, Yi He, Richard Chapman, Hongyi Wu

Computer Science Faculty Publications

Graph neural networks (GNNs) have enabled the automation of many web applications that entail node classification on graphs, such as scam detection in social media and event prediction in service networks. Nevertheless, recent studies revealed that the GNNs are vulnerable to adversarial attacks, where feeding GNNs with poisoned data at training time can lead them to yield catastrophically devastative test accuracy. This finding heats up the frontier of attacks and defenses against GNNs. However, the prior studies mainly posit that the adversaries can enjoy free access to manipulate the original graph, while obtaining such access could be too costly in …

Ready Raider One: Exploring The Misuse Of Cloud Gaming Services, Guannan Liu, Daiping Liu, Shuai Hao, Xing Gao, Kun Sun, Haining Wang

Computer Science Faculty Publications

Cloud gaming has become an emerging computing paradigm in recent years, allowing computer games to offload complex graphics and logic computation to the cloud. To deliver a smooth and high-quality gaming experience, cloud gaming services have invested abundant computing resources in the cloud, including adequate CPUs, top-tier GPUs, and high-bandwidth Internet connections. Unfortunately, the abundant computing resources offered by cloud gaming are vulnerable to misuse and exploitation for malicious purposes. In this paper, we present an in-depth study on security vulnerabilities in cloud gaming services. Specifically, we reveal that adversaries can purposely inject malicious programs/URLs into the cloud gaming services …

Sec-Lib: Protecting Scholarly Digital Libraries From Infected Papers Using Active Machine Learning Framework, Nir Nissim, Aviad Cohen, Jian Wu, Andrea Lanzi, Lior Rokach, Yuval Elovici, Lee Giles

Computer Science Faculty Publications

Researchers from academia and the corporate-sector rely on scholarly digital libraries to access articles. Attackers take advantage of innocent users who consider the articles' files safe and thus open PDF-files with little concern. In addition, researchers consider scholarly libraries a reliable, trusted, and untainted corpus of papers. For these reasons, scholarly digital libraries are an attractive-target and inadvertently support the proliferation of cyber-attacks launched via malicious PDF-files. In this study, we present related vulnerabilities and malware distribution approaches that exploit the vulnerabilities of scholarly digital libraries. We evaluated over two-million scholarly papers in the CiteSeerX library and found the library …

Privacy In Iot Cloud, Aftab Ahmad, Ravi Mukkamala, Karthik Navuluri

Computer Science Faculty Publications

We present a framework for privacy preservation in an information cloud of IoT devices. We contend that privacy provisioning should be located in the user device and must protect the user, the information, and the device from breaches in privacy. We elaborate on how the layered privacy model can ensure such privacy provisioning, and justify the device being the provisioning point instead of the cloud alone. We present the point of view that, due to resource limitations of the IoT devices in general, the privacy preserving measures need to be hard-coded in the device technology. We fall short of suggesting …

An Immersive Telepresence System Using Rgb-D Sensors And Head-Mounted Display, Xinzhong Lu, Ju Shen, Saverio Perugini, Jianjun Yang

Computer Science Faculty Publications

We present a tele-immersive system that enables people to interact with each other in a virtual world using body gestures in addition to verbal communication. Beyond the obvious applications, including general online conversations and gaming, we hypothesize that our proposed system would be particularly beneficial to education by offering rich visual contents and interactivity. One distinct feature is the integration of egocentric pose recognition that allows participants to use their gestures to demonstrate and manipulate virtual objects simultaneously. This functionality enables the instructor to effectively and efficiently explain and illustrate complex concepts or sophisticated problems in an intuitive manner. The …

Automatic Video Self Modeling For Voice Disorder, Ju Shen, Changpeng Ti, Anusha Raghunathan, Sen-Ching S. Cheung, Rita Patel

Computer Science Faculty Publications

Video self modeling (VSM) is a behavioral intervention technique in which a learner models a target behavior by watching a video of him- or herself. In the field of speech language pathology, the approach of VSM has been successfully used for treatment of language in children with Autism and in individuals with fluency disorder of stuttering. Technical challenges remain in creating VSM contents that depict previously unseen behaviors. In this paper, we propose a novel system that synthesizes new video sequences for VSM treatment of patients with voice disorders. Starting with a video recording of a voice-disorder patient, the proposed …

Compression Of Video Tracking And Bandwidth Balancing Routing In Wireless Multimedia Sensor Networks, Yin Wang, Jianjun Yang, Ju Shen, Bryson Payne, Juan Guo, Kun Hua

Computer Science Faculty Publications

There has been a tremendous growth in multimedia applications over wireless networks. Wireless Multimedia Sensor Networks(WMSNs) have become the premier choice in many research communities and industry. Many state-of-art applications, such as surveillance, traffic monitoring, and remote heath care are essentially video tracking and transmission in WMSNs. The transmission speed is constrained by the big file size of video data and fixed bandwidth allocation in constant routing paths. In this paper, we present a CamShift based algorithm to compress the tracking of videos. Then we propose a bandwidth balancing strategy in which each sensor node is able to dynamically select …

Leading Undergraduate Students To Big Data Generation, Jianjun Yang, Ju Shen

Computer Science Faculty Publications

People are facing a flood of data today. Data are being collected at unprecedented scale in many areas, such as networking, image processing, virtualization, scientific computation, and algorithms. The huge data nowadays are called Big Data. Big data is an all encompassing term for any collection of data sets so large and complex that it becomes difficult to process them using traditional data processing applications. In this article, the authors present a unique way which uses network simulator and tools of image processing to train students abilities to learn, analyze, manipulate, and apply Big Data. Thus they develop students hands-on …

Hole Detection And Shape-Free Representation And Double Landmarks Based Geographic Routing In Wireless Sensor Networks, Jianjun Yang, Zongming Fei, Ju Shen

Computer Science Faculty Publications

In wireless sensor networks, an important issue of geographic routing is “local minimum” problem, which is caused by a “hole” that blocks the greedy forwarding process. Existing geographic routing algorithms use perimeter routing strategies to find a long detour path when such a situation occurs. To avoid the long detour path, recent research focuses on detecting the hole in advance, then the nodes located on the boundary of the hole advertise the hole information to the nodes near the hole. Hence the long detour path can be avoided in future routing. We propose a heuristic hole detecting algorithm which identifies …

Structure Preserving Large Imagery Reconstruction, Ju Shen, Jianjun Yang, Sami Taha Abu Sneineh, Bryson Payne, Markus Hitz

Computer Science Faculty Publications

With the explosive growth of web-based cameras and mobile devices, billions of photographs are uploaded to the internet. We can trivially collect a huge number of photo streams for various goals, such as image clustering, 3D scene reconstruction, and other big data applications. However, such tasks are not easy due to the fact the retrieved photos can have large variations in their view perspectives, resolutions, lighting, noises, and distortions. Furthermore, with the occlusion of unexpected objects like people, vehicles, it is even more challenging to find feature correspondences and reconstruct realistic scenes. In this paper, we propose a structure-based image …

Automatic Objects Removal For Scene Completion, Jianjun Yang, Yin Wang, Honggang Wang, Kun Hua, Wei Wang, Ju Shen

Computer Science Faculty Publications

With the explosive growth of Web-based cameras and mobile devices, billions of photographs are uploaded to the Internet. We can trivially collect a huge number of photo streams for various goals, such as 3D scene reconstruction and other big data applications. However, this is not an easy task due to the fact the retrieved photos are neither aligned nor calibrated. Furthermore, with the occlusion of unexpected foreground objects like people, vehicles, it is even more challenging to find feature correspondences and reconstruct realistic scenes. In this paper, we propose a structure-based image completion algorithm for object removal that produces visually …

A Robust Rgbd Slam System For 3d Environment With Planar Surfaces, Po-Chang Su, Ju Shen, Sen-Ching S. Cheung

Computer Science Faculty Publications

With the increasing popularity of RGB-depth (RGB-D) sensors such as the Microsoft Kinect, there have been much research on capturing and reconstructing 3D environments using a movable RGB-D sensor. The key process behind these kinds of simultaneous location and mapping (SLAM) systems is the iterative closest point or ICP algorithm, which is an iterative algorithm that can estimate the rigid movement of the camera based on the captured 3D point clouds. While ICP is a well-studied algorithm, it is problematic when it is used in scanning large planar regions such as wall surfaces in a room. The lack of depth …

Warcreate And Wail: Warc, Wayback, And Heritrix Made Easy, Mat Kelly, Michael L. Nelson, Michele C. Weigle

Computer Science Faculty Publications

[First slide]

The Problem

Institutional Tools, Personal Archivists

• ON YOUR MACHINE

-Complex to Operate

-Require Infrastructure

• DELEGATED TO INSTITUTIONS

-$$$

-Lose original perspective

• Locale content tailoring (DC vs. San Francisco)
• Observation Medium (PC web browser vs. Crawler)

Stochastic Analysis Of Horizontal Ip Scanning, Derek Leonard, Zhongmei Yao, Xiaoming Wang, Dmitri Loguinov

Computer Science Faculty Publications

Intrusion Detection Systems (IDS) have become ubiquitous in the defense against virus outbreaks, malicious exploits of OS vulnerabilities, and botnet proliferation. As attackers frequently rely on host scanning for reconnaissance leading to penetration, IDS is often tasked with detecting scans and preventing them. However, it is currently unknown how likely an IDS is to detect a given Internet-wide scan pattern and whether there exist sufficiently fast scan techniques that can remain virtually undetectable at large-scale. To address these questions, we propose a simple analytical model for the window-expiration rules of popular IDS tools (i.e., Snort and Bro) and utilize a …

Warcreate - Create Wayback-Consumable Warc Files From Any Webpage, Mat Kelly, Michele C. Weigle, Michael L. Nelson

Computer Science Faculty Publications

[First Slide]

What is WARCreate?

• Google Chrome extension
• Creates WARC files
• Enables preservation by users from their browser
• First steps in bringing Institutional Archiving facilities to the PC

Automatic Content Generation For Video Self Modeling, Ju Shen, Anusha Raghunathan, Sen-Ching S. Cheung, Ravi R. Patel

Computer Science Faculty Publications

Video self modeling (VSM) is a behavioral intervention technique in which a learner models a target behavior by watching a video of him or herself. Its effectiveness in rehabilitation and education has been repeatedly demonstrated but technical challenges remain in creating video contents that depict previously unseen behaviors. In this paper, we propose a novel system that re-renders new talking-head sequences suitable to be used for VSM treatment of patients with voice disorder. After the raw footage is captured, a new speech track is either synthesized using text-to-speech or selected based on voice similarity from a database of clean speeches. …

Program Transformations For Information Personalization, Saverio Perugini, Naren Ramakrishnan

Computer Science Faculty Publications

Personalization constitutes the mechanisms necessary to automatically customize information content, structure, and presentation to the end user to reduce information overload. Unlike traditional approaches to personalization, the central theme of our approach is to model a website as a program and conduct website transformation for personalization by program transformation (e.g., partial evaluation, program slicing). The goal of this paper is study personalization through a program transformation lens and develop a formal model, based on program transformations, for personalized interaction with hierarchical hypermedia. The specific research issues addressed involve identifying and developing program representations and transformations suitable for classes of hierarchical …