Digital Commons Network^™

Backdoor Token Unlearning: Exposing And Defending Backdoors In Pretrained Language Models, Peihai Jiang, Xixiang Lyu, Yige Li, Jing Ma

Research Collection School Of Computing and Information Systems

Supervised fine-tuning has become the predominant method for adapting large pretrained models to downstream tasks. However, recent studies have revealed that these models are vulnerable to backdoor attacks, where even a small number of malicious samples can successfully embed backdoor triggers into the model. While most existing defense methods focus on post-training backdoor defense, efficiently defending against backdoor attacks during training phase remains largely unexplored. To address this gap, we propose a novel defense method called Backdoor Token Unlearning (BTU), which proactively detects and neutralizes trigger tokens during the training stage. Our work is based on two key findings: 1) …

Sticky Information Technology Investment: Theory And Empirical Evidence, Peng Liang, Hasan Cavusoglu, Nan Hu

Research Collection School Of Computing and Information Systems

This article provides a new way of thinking about managerial discretion in information technology (IT) investment decisions. We delve into the existence, antecedents, and consequences of sticky IT investment behavior, an understudied managerial deliberate resource commitment decision in response to changes in sales. Guided by downsizing theory, we initially theorize and find that IT investments exhibit stickiness: IT investments move downward less for sales decreases than they move upward for equivalent increases. Then drawing upon agency theory, adjustment costs theory, and managerial expectations theory—which influence managers’ motivation for downsizing—we predict and demonstrate that managers’ empire-building incentives, their avoidance of adjustment …

Deepvmunprotect: Neural Network-Based Recovery Of Vm‑Protected Android Apps For Semantics‑Aware Malware Detection, Xin Zhao, Mu Zhang, Xiaopeng Ke, Yu Pan, Yue Duan, Sheng Zhong, Fengyuan Xu

Research Collection School Of Computing and Information Systems

The emerging virtual machine-based Android packers render existing unpacking techniques ineffective. The state-of-the-art unpacker falls short because it relies on unreliable heuristics and manually crafted semantic models. Hence, it cannot precisely recover app semantics necessary for malware detection. In this paper, we propose DeepVMUnProtect, a deep learning-based approach to automatically and accurately capture the semantics of VM-packed code, so as to facilitate semantic-based Android malware classification. Experiments have shown that DeepVMUnProtect outperforms the state-of-the-art tool on recovering opcode semantics in Qihoo(58.3%), Baidu(47.5%) and NMMP (58.8%) respectively, and can enable semantics-aware malware detection which prior work fails to do.

Exploring Intelligent Manufacturing: How Artificial Intelligence Affects Productivity And Labor Demand At The Enterprise Level, Jie Gu

Dissertations and Theses Collection (Open Access)

Manufacturing is a cornerstone of national economic health and social stability, yet it faces challenges such as declining profits, rising labor costs, and an aging workforce. In China, the manufacturing sector is undergoing a critical transformation, driven by technological advancements like artificial intelligence (AI) and the push for intelligent manufacturing. This study explores how AI revitalizes the manufacturing sector by enhancing enterprise productivity and reshaping labor demand, with a focus on quality inspection processes. Using a leading bearing factory as a case study, the research employs econometric models, A/B testing, and interviews to quantify AI’s impact on production efficiency, costs, …

Adversarial Attacks And Defense Methods In Robotic Systems, Thanh D. Le

Shelby Hall Graduate Research Forum Presentations

Presentation slides for a presentation given at the 1st annual Shelby Hall Graduate Research Forum at the University of South Alabama.

Bpen: Brain Posterior Evidential Network For Trustworthy Brain Imaging Analysis, Kai Ye, Haoteng Tang, Siyuan Dai, Igor Fortel, Paul M. Thompson, R. Scott Mackin, Alex Leow, Heng Huang, Liang Zhan

Computer Science Faculty Publications

The application of deep learning techniques to analyze brain functional magnetic resonance imaging (fMRI) data has led to significant advancements in identifying prospective biomarkers associated with various clinical phenotypes and neurological conditions. Despite these achievements, the aspect of prediction uncertainty has been relatively underexplored in brain fMRI data analysis. Accurate uncertainty estimation is essential for trustworthy learning, given the challenges associated with brain fMRI data acquisition and the potential diagnostic implications for patients. To address this gap, we introduce a novel posterior evidential network, named the Brain Posterior Evidential Network (BPEN), designed to capture both aleatoric and epistemic uncertainty in …

Private Reachability Queries On Structured Encrypted Temporal Bipartite Graphs, Yulin Wu, Lanxiang Chen, Gaolin Chen, Yi Mu, Robert H. Deng

Research Collection School Of Computing and Information Systems

A temporal bipartite graph is a graph model that incorporates time-related information into its edges, making it suitable for modeling real-world phenomena like disease outbreaks. However, this temporal information is often sensitive. To protect the privacy of graph data, researchers have explored various approaches to preserve privacy in graph queries, with reachability queries being popular and fundamental as they determine the possibility of reaching one node from others in a graph. While privacy-preserving reachability queries have been extensively studied, existing efforts often overlook the valuable attribute information present in both edges and nodes of the graphs. Moreover, reachability queries on …

Exploring Ai Technology In Grammar Performance Testing For Children With Learning Disabilities, Dimitra V. Katsarou, Evangelos Mantsos, Soultana Papadopoulou, Maria Sofologi, Efthymia Efthymiou, Ilias Vasileiou, Kalliopi Megari, Maria Theodoratou, Georgios A. Kougioumtzis

All Works

The study explores the application of artificial intelligence (AI) in addressing grammar challenges among children with learning disabilities, aiming to assess the efficacy of an AI-driven tool for personalized interventions. A sample of 100 children aged 8–12, diagnosed with learning disabilities, was recruited from special education programs. Participants were divided into an experimental group (n = 50), which used an AI-based grammar assessment tool with personalized feedback, and a control group (n = 50), which completed conventional paper-based grammar tests without feedback. The AI tool administered adaptive grammar tasks, including sentence correction and verb conjugation, and performance was evaluated over …

Identifying Human Factor Causes Of Remotely Piloted Aircraft System Safety Occurrences In Australia, John Murray, Steven Richardson, Keith Joiner, Graham Wild

Research outputs 2022 to 2026

Remotely piloted aircraft are a fast-emerging sector of the aviation industry. Although technical failures have been the largest cause of accident occurrences for Remotely Piloted Aircraft Systems (RPASs), if they are to follow the path of conventionally crewed aviation, Human Factors (HFs) will increasingly contribute to accidents as the technology of RPASs improves. Examining an RPAS accident database from 2008–2019 for HF-caused accidents and coding to the Human Factors Analysis and Classification System (HFACS) taxonomy, an exploration of RPAS HFs is carried out and the predominant HF issues for RPAS pilots identified. The majority of HF accidents were coded to …

Pathways To Chronic Disease Detection And Prediction: Mapping The Potential Of Machine Learning To The Pathophysiological Processes While Navigating Ethical Challenges, Ebenezer Afrifa-Yamoah, Eric Adua, Emmanuel Peprah-Yamoah, Enoch O. Anto, Victor Opoku-Yamoah, Emmanuel Acheampong, Michael J. Macartney, Rashid Hashmi

Research outputs 2022 to 2026

Chronic diseases such as heart disease, cancer, and diabetes are leading drivers of mortality worldwide, underscoring the need for improved efforts around early detection and prediction. The pathophysiology and management of chronic diseases have benefitted from emerging fields in molecular biology like genomics, transcriptomics, proteomics, glycomics, and lipidomics. The complex biomarker and mechanistic data from these “omics” studies present analytical and interpretive challenges, especially for traditional statistical methods. Machine learning (ML) techniques offer considerable promise in unlocking new pathways for data-driven chronic disease risk assessment and prognosis. This review provides a comprehensive overview of state-of-the-art applications of ML algorithms for …

Forward-Secure Hierarchical Delegable Signature For Smart Homes, Jianfei Sun, Guowen Xu, Yang Yang, Xuehuan Yang, Xiaoguo Li, Cong Wu, Zhen Liu, Guomin Yang, Robert H. Deng

Research Collection School Of Computing and Information Systems

Aiming to provide people with great convenience and comfort, smart home systems have been deployed in thousands of homes. In this paper, we focus on handling the security and privacy issues in such a promising system by customizing a new cryptographic primitive to provide the following security guarantees: (1) fine-grained, privacy-preserving authorization for smart home users and integrity protection of communication contents; (2) flexible self-sovereign permission delegation; (3) forward security of previous messages. To our knowledge, no previous system has been designed to consider these three security and privacy requirements simultaneously. To tackle these challenges, we put forward the first-ever …

Artificial Intelligence In Surgical Coding: Evaluating Large Language Models For Current Procedural Terminology Accuracy In Hand Surgery, Emily Isch, Jamie Lee, D. Mitchell Self, Abhijeet Sambangi, Theodore E. Habarth-Morales, John R. Vaile, E. J. Caterson

Department of Surgery Faculty Papers

PURPOSE: The advent of large language models (LLMs) like ChatGPT has introduced notable advancements in various surgical disciplines. These developments have led to an increased interest in the use of LLMs for Current Procedural Terminology (CPT) coding in surgery. With CPT coding being a complex and time-consuming process, often exacerbated by the scarcity of professional coders, there is a pressing need for innovative solutions to enhance coding efficiency and accuracy.

METHODS: This observational study evaluated the effectiveness of five publicly available large language models-Perplexity.AI, Bard, BingAI, ChatGPT 3.5, and ChatGPT 4.0-in accurately identifying CPT codes for hand surgery procedures. A …

Evaluating Learning Outcomes In A Serious Game: A Practical And Model Checking Approach, Matthew D. Douglas

Theses and Dissertations

This research introduces a novel computational framework to evaluate and predict the educational impact of serious games during development. By using finite state machines (FSM) and model-checking techniques, this study evaluates two serious games. Traditional evaluation approaches, often reliant on resource-intensive human trials, lack scalability and fail to provide early insight into the alignment of game mechanics with learning objectives. This study addresses these challenges of traditional evaluation methods.

Graph Neural Network-Based Uav Coverage Planning For Robust And Efficient 3d Environments, Gal Tsfaty

Theses and Dissertations

This thesis addresses the challenge of generating optimized UAV waypoints for complete coverage of complex 3D environments, utilizing graph-based computational techniques. The proposed framework replaces computationally intensive steps—triangulation and three-coloring—within the Vantage Waypoint Set Generation Algorithm (VWSGA) pipeline with Graph Neural Networks (GNNs). By learning structural patterns, the GNN achieves scalable and robust triangulation and node classification, enabling enhanced coverage planning in irregular geometries. A novel penalty mechanism ensures alignment with graph structure during adjacency prediction. Experimental results demonstrate the effectiveness of GNNs in balancing accuracy, computational efficiency, and adaptability, advancing UAV coverage optimization.

Standardizing Canine Breed Data In Veterinary Records Is Challenging, But Computer Vision Offers An Alternative Perspective On Breed Assignment, Glenvelis Perez, Yixuan He, Zihan Lyu, Yilin Chen, Nicholas Howe, Halie M. Rando

Computer Science: Faculty Publications

Dog breed is fundamental health information, especially in the context of breed-linked diseases. The standard-ization of breed terminology across health records is necessary to leverage the big data revolution for veterinary research. Breed can also inform clinical decision making. However, client-reported breeds vary in their reliability depending on how breed was determined. Surprisingly, research in computer science reports that AI can assign breed to dogs with over 90% accuracy from a photograph. Here, we explore the extent to which current research in AI is relevant to breed assignment or validation in veterinary contexts. This review provides a primer on approaches …

Performing Requirements Specification And Analysis Through Open Generative Pre-Trained Transformers, Harvey J. Hurst

Theses and Dissertations

Every acquisition program begins with a requirement, and for those programs to succeed, robust requirements engineering (RE) must be implemented. RE encompasses eliciting, analyzing, specifying, and validating requirements—a critical process throughout a program's lifecycle. Despite its importance, RE faces challenges such as scope creep, ambiguity, redundancy, and inadequate automation support, often exacerbated by reliance on historical data. To address these issues, this thesis leverages advancements in Generative Technology, particularly large language models (LLMs) such as Generative Pre-Trained Transformers (GPTs). This research developed two GPT-based tools: the Single Requirement Analysis Tool and the Set of Requirements Analysis Tool. These tools were …

A Machine Learning/Deep Learning Investigation On Remote Manufacturing Machine State Classification, Ajeet S. Parmar

Theses and Dissertations

Determining the extent of manufacturing capabilities with respect to adversarial or hostile nations is a topic of significant importance to the Department of Defense. Manufacturing capabilities can serve as indications of a nation's industrial power and its economy of force in warfare. Remotely detecting machine operations via electromagnetic sensors may be possible via Deep Learning (DL) and Machine Learning (ML) algorithms. To predict machine states, sensor data is collected externally from a machine shop on a college campus to monitor the operating states of lathes and mills in individual and concurrent operation. Furthermore, several sensors are placed in various positions, …

Imageinthat: Manipulating Images To Convey User Instructions To Robots, Karthik Mahadevan, Blaine Lewis, Jiannan Li, Bilge Mutlu, Anthony Tang, Tovi Grossman

Research Collection School Of Computing and Information Systems

Foundation models are rapidly improving the capability of robots in performing everyday tasks autonomously such as meal preparation, yet robots will still need to be instructed by humans due to model performance, the difficulty of capturing user preferences, and the need for user agency. Robots can be instructed using various methods---natural language conveys immediate instructions but can be abstract or ambiguous, whereas end-user programming supports longer-horizon tasks but interfaces face difficulties in capturing user intent. In this work, we propose using direct manipulation of images as an alternative paradigm to instruct robots, and introduce a specific instantiation called ImageInThat which …

Explainable Neural Networks With Guarantees: A Sparse Estimation Approach, Antoine Ledent, Peng Liu

Research Collection School Of Computing and Information Systems

Balancing predictive power and interpretability has long been a challenging research area, particularly in powerful yet complex models like neural networks, where nonlinearity obstructs direct interpretation. This paper introduces a novel approach to constructing an explainable neural network that harmonizes predictiveness and explainability. Our model is designed as a linear combination of a sparse set of jointly learned features, each derived from a different trainable function applied to a single 1-dimensional input feature. Leveraging the ability to learn arbitrarily complex relationships, our neural network architecture enables automatic selection of a sparse set of important features, with the final prediction being …

Multisfl: Towards Accurate Split Federated Learning Via Multi-Model Aggregation And Knowledge Replay, Zeke Xia, Ming Hu, Dengke Yan, Ruixuan Liu, Anran Li, Xiaofei Xie, Mingsong Chen

Research Collection School Of Computing and Information Systems

Although Split Federated Learning (SFL) effectively enables knowledge sharing among resource-constrained clients, it suffers from low training performance due to the neglect of data heterogeneity and catastrophic forgetting problems. To address these issues, we propose a novel SFL approach named MultiSFL, which adopts i) an effective multimodel aggregation mechanism to alleviate gradient divergence caused by heterogeneous data and ii) a novel knowledge replay strategy to deal with the catastrophic forgetting problem. MultiSFL adopts two servers (i.e., the fed server and main server) to maintain multiple branch models for local training and an aggregated master model for knowledge sharing among branch …

Lightprof: A Lightweight Reasoning Framework For Large Language Model On Knowledge Graph, Tu Ao, Yanhua Yu, Yuling Wang, Yang Deng, Zirui Guo, Liang Pang, Pinghui Wang, Tat-Seng Chua, Xiao Zhang, Zhen Cai

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) have impressive capabilities in text understanding and zero-shot reasoning. However, delays in knowledge updates may cause them to reason incorrectly or produce harmful results. Knowledge Graphs (KGs) provide rich and reliable contextual information for the reasoning process of LLMs by structurally organizing and connecting a wide range of entities and relations. Existing KG-based LLM reasoning methods only inject KGs’ knowledge into prompts in a textual form, ignoring its structural information. Moreover, they mostly rely on close-source models or open-source models with large parameters, which poses challenges to high resource consumption. To address this, we propose a …

How To Securely Delegate And Revoke Partial Authorization Credentials, Meng Sun, Junzuo Lai, Wei Wu, Ye Yang, Cheng-Kang Chu, Robert H. Deng

Research Collection School Of Computing and Information Systems

An attribute-based credential (ABC) system allows a user, obtaining a credential on a set of attributes from an issuer, to anonymously prove a subset of attributes to a service provider. Nowadays, delegation is an important requirement of ABC, which allows a user to delegate his credentials to other users. However, traditional delegatable ABC systems only support delegating a credential with all attributes. In many scenarios, an appropriate delegation is a user can delegate his credential on parts of attributes to others. Another requirement is revocation of credentials in case of unexpected events. In this article, we propose a delegatable and …

Cachealarm: Monitoring Sensitive Behaviors Of Android Apps Using Cache Side Channel, Jianwen Tian, Haoyu Ma, Debin Gao, Xiaohui Kuang

Research Collection School Of Computing and Information Systems

Malware attack has been a serious threat to the security and privacy of both individual and corporation users of the Android platform. Business entities seek to protect themselves by means of monitoring privacy-related sensitive behaviors conducted on company-issued Android devices. However, due to Android’s own access control and privacy protection policies, this is difficult to be done with third-party apps using only normal privileges. Existing works proposed using side-channel readings from leaky APIs and system virtual files to speculate runtime app behaviors, which could be unreliable due to future system updates (that ban exploited resources), hardware jittering, etc. In this …

Privacy-Preserving Fine-Grained Data Sharing With Dynamic Service For The Cloud-Edge Iot, Jianfei Sun, Yangyang Bao, Weidong Qiu, Rongxing Lu, Songnian Zhang, Yunguo Guan, Xiaochun Cheng

Research Collection School Of Computing and Information Systems

The cloud-edge computing model has been expected to play a revolutionary role in promoting the quality of future generation large-scale Internet of Things (IoT) services. However, security and privacy in data sharing remain crucial issues hindering the success of cloud-edge IoT services. While some solutions based on attribute-based encryption (ABE) have been proposed to address these issues, they still face practical challenges such as attribute privacy leakage, resource-constrained devices, dynamic user groups, inflexible and inefficient service response. To address these challenges, this paper proposes a privacy-preserving fine-grained data sharing scheme with dynamic service (PF2DS), which implements access control by calculating …

Enhanced Sample Selection With Confidence Tracking: Identifying Correctly Labeled Yet Hard-To-Learn Samples In Noisy Data, Weiran Pan, Wei Wei, Feida Zhu, Yong Deng

Research Collection School Of Computing and Information Systems

We propose a novel sample selection method for image classification in the presence of noisy labels. Existing methods typically consider small-loss samples as correctly labeled. However, some correctly labeled samples are inherently difficult for the model to learn and can exhibit high loss similar to mislabeled samples in the early stages of training. Consequently, setting a threshold on per-sample loss to select correct labels results in a trade-off between precision and recall in sample selection: a lower threshold may miss many correctly labeled hard-to-learn samples (low recall), while a higher threshold may include many mislabeled samples (low precision). To address …

Adversarial Attacks On Event-Based Pedestrian Detectors: A Physical Approach, Guixu Lin, Muyao Niu, Qingtian Zhu, Zhengwei Yin, Zhuoxiao Li, Shengfeng He, Yinqiang Zheng

Research Collection School Of Computing and Information Systems

Event cameras, known for their low latency and high dynamic range, show great potential in pedestrian detection applications. However, while recent research has primarily focused on improving detection accuracy, the robustness of event-based visual models against physical adversarial attacks has received limited attention. For example, adversarial physical objects, such as specific clothing patterns or accessories, can exploit inherent vulnerabilities in these systems, leading to misdetections or misclassifications. This study is the first to explore physical adversarial attacks on event-driven pedestrian detectors, specifically investigating whether certain clothing patterns worn by pedestrians can cause these detectors to fail, effectively rendering them unable …

Simulation-Free Hierarchical Latent Policy Planning For Proactive Dialogues, Tao He, Lizi Liao, Yixin Cao, Yuanxing Liu, Yiheng Sun, Zerui Chen, Ming Liu, Bing Qin

Research Collection School Of Computing and Information Systems

Recent advancements in proactive dialogues have garnered significant attention, particularly for more complex objectives (e.g. emotion support and persuasion). Unlike traditional task-oriented dialogues, proactive dialogues demand advanced policy planning and adaptability, requiring rich scenarios and comprehensive policy repositories to develop such systems. However, existing approaches tend to rely on Large Language Models (LLMs) for user simulation and online learning, leading to biases that diverge from realistic scenarios and result in suboptimal efficiency. Moreover, these methods depend on manually defined, context-independent, coarse-grained policies, which not only incur high expert costs but also raise concerns regarding their completeness. In our work, we …

Evaluating Software Development Agents: Patch Patterns, Code Quality, And Issue Complexity In Real-World Github Scenarios, Zhi Chen, Lingxiao Jiang

Research Collection School Of Computing and Information Systems

In recent years, AI-based software engineering has progressed from pre-trained models to advanced agentic workflows, with Software Development Agents representing the next major leap. These agents, capable of reasoning, planning, and interacting with external environments, offer promising solutions to complex software engineering tasks. However, while much research has evaluated code generated by large language models (LLMs), comprehensive studies on agent-generated patches, particularly in real-world settings, are lacking. This study addresses that gap by evaluating 4,892 patches from 10 top-ranked agents on 500 real-world GitHub issues from SWE-Bench Verified, focusing on their impact on code quality. Our analysis shows no single …

Learning To Identify Seen, Unseen And Unknown In The Open World: A Practical Setting For Zero-Shot Learning, Sethupathy Parameswaran, Yuan Fang, Chandan Gautam, Savitha Ramasamy, Xiaoli Li

Research Collection School Of Computing and Information Systems

As vision-language models advance, addressing the Zero-Shot Learning (ZSL) problem in the open world becomes increasingly crucial. Specifically, a robust model must handle three types of samples during inference: seen classes with visual and semantic information provided in training, unseen classes with only the semantic information in training, and unknown samples with no prior information from training. Existing methods either handle seen and unseen classes together (ZSL) or seen and unknown classes (known as Open-Set Recognition, OSR). However, none addresses the simultaneous handling of all three, which we term Open-Set Zero-Shot Learning (OZSL). To address this problem, we propose a …

A Heterogeneous Graph-Based Multi-Task Learning For Fault Event Diagnosis In Smart Grid, Dibaloke Chanda, Nasim Yahyasoltani

Computer Science Faculty Research and Publications

Precise and timely fault diagnosis is a prerequisite for a distribution system to ensure minimum downtime and maintain reliable operation. This necessitates access to a comprehensive procedure that can provide the grid operators with insightful information in the case of a fault event. In this paper, we propose a heterogeneous multi-task learning graph neural network (MTL-GNN) capable of detecting, locating and classifying faults in addition to providing an estimate of the fault resistance and current. Using a graph neural network (GNN) allows for learning the topological representation of the distribution system as well as feature learning through a message-passing scheme. …