Digital Commons Network^™

Interpretable And Robust Deep Anomaly Detection, He Cheng

All Graduate Theses and Dissertations, Fall 2023 to Present

Anomaly detection is crucial in fields like cybersecurity, healthcare, and finance, as it helps identify unusual or potentially harmful events in data. With the rise of deep learning, advanced models have been developed for anomaly detection, but they often operate as "black boxes" that lack transparency and can be susceptible to malicious attacks. My research addresses these issues by creating methods that make deep learning-based anomaly detection more understandable and by investigating how such models can be compromised by backdoor attacks.

To improve transparency, I propose three methods that explain how these models detect anomalies. The first method, called Anomalous …

Dynamic Key-Based Privacy-Preserving Authentication Scheme For Internet Of Drones, Zain Chaudhary

Honors Theses

The Internet of Drones (IoD) proliferation has catalyzed transformative changes across various industries, from agriculture to urban management. However, expanding drone networks also presents significant security challenges concerning secure communication and authentication. This paper introduces a robust privacy-preserving key-based authentication scheme tailored explicitly for the IoD, utilizing a matrix key generated by Hierarchical Message Authentication Codes (HMAC) and the SHA-256 algorithm to address these vulnerabilities. Our system enhances security by ensuring each drone in the network can authenticate securely and reliably with a central unit, preventing unauthorized access and securing communications against common threats like eavesdropping and impersonation attacks. Our …

Human Vs. Ai Counseling: College Students' Perspectives, Mohammad Amin Kuhail, Nazik Alturki, Justin Thomas, Amal K. Alkhalifa

All Works

Transitioning to college life while navigating the complexities of emerging adulthood can be stressful. In some instances, it may even lead to the onset of mental health problems or the exacerbation of existing issues. While therapeutic resources are typically available in tertiary educational contexts, social stigma may lead to service underutilization. Additionally, high student-to-therapist ratios can create bottlenecks to access when such services are sought. Offering an adjunct to traditional campus counseling services, AI chatbots can potentially address such issues. Chatbots can provide flexible, accessible, anonymous, and cost-effective first-line support, improving access and extending traditional treatment methodologies. This study evaluates …

Exploring The Capabilities Of Classifier-Free Guidance In Recommendation Tasks, Noah Buchanan

Graduate Theses and Dissertations

This thesis addresses the problem of recommending items to users based on their ratings of items through a diffusion recommender system. Regular recommender systems are already capable of efficient recommendation through conventional methods such as collaborative or content-based filtering. Diffusion is a new type of generative AI that aims to improve our previous AI's shortcomings in the generative domain, like Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs). We use diffusion to create a recommender system that mirrors the sequence users take when browsing and rating items. The current methods of recommendation with diffusion do not use the new innovation …

Comparing And Evaluating Models Of Tile-Assembly Using Intrinsic Simulation, Daniel Hader

Graduate Theses and Dissertations

Tile-assembly studies abstract models of computation inspired by advancements in the emerging field of DNA-nanotechnology, where synthetic strands of DNA are used as building blocks for microscopic structures. These DNA strands can be made to combine into structural units with selectively sticky sides that abstractly resemble Wang tiles. However, unlike Wang tiles, an assembly process is modeled where tiles combine one-by-one to form larger assemblies according to matching rules based on their sticky sides. While in practice, these tile-assembly models have seen use in designing DNA nano-structures, the mathematical study of their theory has revealed an exciting interplay between geometric …

Addressing Cybersecurity Data & Workforce Scarcity With Troy: Testbed For Resilient Operational Systems, Henry Oliver Schmidt

Graduate Theses and Dissertations

Machine learning has seen an explosive rise in the past decade. Companies, organizations, and governments are racing to pursue the advancements and insight provided by machine learning powered tools. However, to get effective and meaningful insights from machine learning models a significant amount of detailed data is required to train them. This poses a problem in fields where data is not openly available, such as cybersecurity. Entities are often unwilling to give out network or system data to the public for machine learning and cybersecurity research since that data can contain sensitive or proprietary information. The risk simply outweighs the …

Enhancing Smart Grid Security And Resilience Using Programmable Networks, Zheng Hu

Graduate Theses and Dissertations

The security and resilience of smart grids are essential to ensuring reliable and efficient energy distribution, especially as these cyber-physical systems grow more interconnected and complex. Supervisory Control and Data Acquisition (SCADA) systems play a critical role in smart grid operations by enabling essential infrastructure control and real-time monitoring. However, SCADA systems are highly vulnerable to modern cyber threats, which target weaknesses in industrial protocols and real-time data requirements.

This dissertation investigates the potential of programmable network technologies, with a focus on P4 (Programming Protocol-independent Packet Processors) switch, to deliver adaptable, in-network security solutions tailored to the needs of smart …

Triadic Temporal-Semantic Alignment For Weakly-Supervised Video Moment Retrieval, Jin Liu, Jialong Xie, Fengyu Zhou, Shengfeng He

Research Collection School Of Computing and Information Systems

Video Moment Retrieval (VMR) aims to identify specific event moments within untrimmed videos based on natural language queries. Existing VMR methods have been criticized for relying heavily on moment annotation bias rather than true multi-modal alignment reasoning. Weakly supervised VMR approaches inherently overcome this issue by training without precise temporal location information. However, they struggle with fine-grained semantic alignment and often yield multiple speculative predictions with prolonged video spans. In this paper, we take a step forward in the context of weakly supervised VMR by proposing a triadic temporalsemantic alignment model. Our proposed approach augments weak supervision by comprehensively addressing …

Towards Unified Multimodal Editing With Enhanced Knowledge Collaboration, Kaihang Pan, Zhaoyu Fan, Juncheng Li, Qifan Yu, Hao Fei, Siliang Tang, Richang Hong, Hanwang Zhang, Qianru Sun

Research Collection School Of Computing and Information Systems

The swift advancement in Multimodal LLMs (MLLMs) also presents significant challenges for effective knowledge editing. Current methods, including intrinsic knowledge editing and external knowledge resorting, each possess strengths and weaknesses, struggling to balance the desired properties of reliability, generality, and locality when applied to MLLMs. In this paper, we propose UniKE, a novel multimodal editing method that establishes a unified perspective and paradigm for intrinsic knowledge editing and external knowledge resorting. Both types of knowledge are conceptualized as vectorized key-value memories, with the corresponding editing processes resembling the assimilation and accommodation phases of human cognition, conducted at the same semantic …

Mvgamba : Unify 3d Content Generation As State Space Sequence Modeling, Xuanyu Yi, Zike Wu, Qiuhong Shen, Qingshan Xu, Pan Zhou, Joo-Hwee Lim, Shuicheng Yan, Xinchao Wang, Hanwang Zhang

Research Collection School Of Computing and Information Systems

Recent 3D large reconstruction models (LRMs) can generate high-quality 3D content in sub-seconds by integrating multi-view diffusion models with scalable multi-view reconstructors. Current works further leverage 3D Gaussian Splatting as 3D representation for improved visual quality and rendering efficiency. However, we observe that existing Gaussian reconstruction models often suffer from multi-view inconsistency and blurred textures. We attribute this to the compromise of multi-view information propagation in favor of adopting powerful yet computationally intensive architectures (e.g., Transformers). To address this issue, we introduce MVGamba, a general and lightweight Gaussian reconstruction model featuring a multi-view Gaussian reconstructor based on the RNN-like State …

Lova3 : Learning To Visual Question Answering, Asking And Assessment, Henry Hengyuan Zhao, Pan Zhou, Difei Gao, Bai Shou, Mike Zheng Shou

Research Collection School Of Computing and Information Systems

Question answering, asking, and assessment are three innate human traits crucial for understanding the world and acquiring knowledge. By enhancing these capabilities, humans can more effectively utilize data, leading to better comprehension and learning outcomes. Current Multimodal Large Language Models (MLLMs) primarily focus on question answering, often neglecting the full potential of questioning and assessment skills. Inspired by the human learning mechanism, we introduce LOVA3 , an innovative framework named “Learning tO Visual question Answering, Asking and Assessment,” designed to equip MLLMs with these additional capabilities. Our approach involves the creation of two supplementary training tasks GenQA and EvalQA, aiming …

Sampdetox : Black-Box Backdoor Defense Via Perturbation-Based Sample Detoxification, Yanxin Yang, Chentao Jia, Dengke Yan, Ming Hu, Tianlin Li, Xiaofei Xie, Xian Wei, Mingsong Chen

Research Collection School Of Computing and Information Systems

The advancement of Machine Learning has enabled the widespread deployment of Machine Learning as a Service (MLaaS) applications. However, the untrustworthy nature of third-party ML services poses backdoor threats. Existing defenses in MLaaS are limited by their reliance on training samples or white-box model analysis, highlighting the need for a black-box backdoor purification method. In our paper, we attempt to use diffusion models for purification by introducing noise in a forward diffusion process to destroy backdoors and recover clean samples through a reverse generative process. However, since a higher noise also destroys the semantics of the original samples, it still …

Sprinql : Sub-Optimal Demonstrations Driven Offline Imitation Learning, Minh Huy Hoang, Tien Mai, Pradeep Varakantham

Research Collection School Of Computing and Information Systems

We focus on offline imitation learning (IL), which aims to mimic an expert's behavior using demonstrations without any interaction with the environment. One of the main challenges in offline IL is the limited support of expert demonstrations, which typically cover only a small fraction of the state-action space. While it may not be feasible to obtain numerous expert demonstrations, it is often possible to gather a larger set of sub-optimal demonstrations. For example, in treatment optimization problems, there are varying levels of doctor treatments available for different chronic conditions. These range from treatment specialists and experienced general practitioners to less …

Self-Supervised Fine-Tuning For Neural Expert Finding, Budhitama Subagdja, Dan Sanchari, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Expert finding systems allow ones to find individuals who have expertise in specific fields or domains. Traditional expert finding are mostly based on topic modeling or keyword search methods that are limited in their capability to encode contextual knowledge from natural language. To address the limitation, this paper presents Neural Expert Finder (NEF), a novel method that takes a transfer learning approach based on transformer encoder networks to leverage the rich seman-tic and syntactic patterns of language encoded in pre-trained language models (PLMs). We propose a self-supervised learning approach utilizing contrastive training using both positive and automatically generated negative samples …

Long-Tailed Out-Of-Distribution Detection Via Normalized Outlier Distribution Adaptation, Wenjun Miao, Guansong Pang, Jin Zheng, Xiao Bai

Research Collection School Of Computing and Information Systems

Onekeychallenge in Out-of-Distribution (OOD) detection is the absence of groundtruth OOD samples during training. One principled approach to address this issue is to use samples from external datasets as outliers (i.e., pseudo OOD samples) to train OOD detectors. However, we find empirically that the outlier samples often present a distribution shift compared to the true OOD samples, especially in LongTailed Recognition (LTR) scenarios, where ID classes are heavily imbalanced, i.e., the true OOD samples exhibit very different probability distribution to the head and tailed ID classes from the outliers. In this work, we propose a novel approach, namely normalized outlier …

Divlog: Log Parsing With Prompt Enhanced In-Context Learning, Junjielong Xu, Ruichun Yang, Yintong Huo, Chengyu Zhang, Pinjia He

Research Collection School Of Computing and Information Systems

Log parsing, which involves log template extraction from semistructured logs to produce structured logs, is the first and the most critical step in automated log analysis. However, current log parsers suffer from limited effectiveness for two reasons. First, traditional data-driven log parsers solely rely on heuristics or handcrafted features designed by domain experts, which may not consistently perform well on logs from diverse systems. Second, existing supervised log parsers require model tuning, which is often limited to fixed training samples and causes sub-optimal performance across the entire log source. To address this limitation, we propose DivLog, an effective log parsing …

Exploring The Cognitive Sense Of Self In Ai: Ethical Frameworks And Technological Advances For Enhanced Decision-Making, Emily Barnes, James Hutson

Faculty Scholarship

The burgeoning field of Artificial Intelligence (AI) increasingly focuses on developing systems capable of self-awareness, merging technological innovation with deep ethical and philosophical considerations. This article explores the cognitive sense of self within AI, examining mechanisms through which AI systems may mirror human-like consciousness and self-perception. Despite significant advances, substantial gaps remain in the understanding and practical implementation of self-aware characteristics in AI, particularly in applying theoretical models and ethical frameworks to real-world scenarios. There is a pressing need for comprehensive research to explore these theoretical underpinnings and translate them into operational systems capable of ethical and adaptable behaviors. This …

An Improved Conditional Integrator Sliding Mode Controller Based On Swarm Intelligence For A Magnetic Levitation System, Abdelkader Kerraci, Mohamed Fayçal Khelfi, Zoubir Ahmed-Foitih

Turkish Journal of Electrical Engineering and Computer Sciences

This paper proposes an enhanced Conditional Integrator Sliding Mode Controller using Particle Swarm Optimization (CISMCPSO) for a magnetic levitation system (MLS). The main advantage of this controller is its robustness to uncertainties and disturbances, which also avoids chattering and ensures zero static steady-state error. The main idea of CISMCPSO is to activate its integral action only when the sliding surface reaches the boundary layer while it is reduced to zero or close to zero elsewhere, which avoids destroying the transient response caused by the conventional integral sliding-mode controller. A particle swarm optimization algorithm schedules the conditional integral term parameter of …

Developing Linguistic Patterns To Mitigate Inherent Human Bias In Offensive Language Detection, Toygar Tanyel, Besher Alkurdi, Serkan Ayvaz

Turkish Journal of Electrical Engineering and Computer Sciences

With the proliferation of social media, there has been a sharp increase in offensive content, particularly targeting vulnerable groups, exacerbating social problems such as hatred, racism, and sexism. Detecting offensive language use is crucial to prevent offensive language from being widely shared on social media. However, the accurate detection of irony, implication, and various forms of hate speech on social media remains a challenge. Natural language-based deep learning models require extensive training with large, comprehensive, and labeled datasets. Unfortunately, manually creating such datasets is both costly and error-prone. Additionally, the presence of human-bias in offensive language datasets is a major …

Fault Diagnosis Of Photovoltaic Array Based On Gated Residual Network With Multi-Head Self Attention Mechanism, Ahmed Mesai Belgacem, Mounir Hadef, Abdesslem Djerdir

Turkish Journal of Electrical Engineering and Computer Sciences

Effective fault identification and diagnosis in photovoltaic (PV) arrays is vital for improving the effectiveness, and safety of solar energy systems. While various artificial intelligence methods have successfully established fault detection and diagnosis models, introducing inefficiencies and potentially overlooking useful features. Moreover, these methods often employ neural networks with limited performance capabilities. In response to these challenges, this paper introduces an innovative intelligent model that integrates a combination of a gated residual neural network (GRN) and a multi-head self-attention mechanism (MHSA). To evaluate the proposed fault diagnosis model, the small-scale PV grid system is implemented, and fault simulation experiments, including …

Development Of A Web-Based Information System For Student Leave Permission At Dar Al-Raudhah Islamic Boarding School: Iso Quality Standards Analysis, Bonita Destiana, Priyanto Priyanto, Rahmatul Irfan, Muhammad Gus Khamim, Muhammad Yusuf Ridlo, Muhammad Iqbal

Elinvo (Electronics, Informatics, and Vocational Education)

Dar Al-Raudhah Entrepreneur, Islamic Boarding School, has adopted digital technology by upgrading hardware and software also investing in reliable internet infrastructure. However, this school still faces issues with students’ leave permission process due to reliance on manual bookkeeping and Excel, which leads to potential errors. Based on those problems, this research aims to create a web-based student leave permission system called SIPERSAN. The SIPERSAN system was developed with a Waterfall development model, which includes requirements analysis, design, implementation, testing, and deployment. The database is managed with MySQL, and the system is developed using PHP with the Laravel framework. Based on …

Relationship-Influenced Cyber Hygiene (Rich) In Community Banks, Monte L. Ward

USF Tampa Graduate Theses and Dissertations

This dissertation is a research study that introduces the theoretical model of Relationship-Influenced Cyber Hygiene (RICH) through its investigation of the phenomenon the researcher experienced: how the interpersonal relationship between top management and cybersecurity personnel within smaller community banks influences the cyber hygiene of top management. Due to the value proposition of community banks to their clients and the limited budget of smaller institutions for investment in cybersecurity controls and initiates, community banks need additional strategies to achieve stronger cyber hygiene, especially as it relates the greatest weakness in most cybersecurity infrastructures—the human element. This study identifies and addresses a …

Designing Customized Loss Functions For Training Deep Neural Networks, Ali Pourramezan Fard

Electronic Theses and Dissertations

This dissertation explores the critical role of loss functions in enhancing the predictive performance of deep machine learning models. Loss functions are an integral element of all the ongoing advances we witness daily in this domain. I design custom loss functions and their impacts on various machine learning tasks, particularly in computer vision.

In the first stage of my research, I aim to improve the prediction performance of deep learning models by providing them with more precise feedback associated with task requirements. This led me to create the concept of assistive loss functions. My first proposed loss function, inspired by …

Applying Machine‐Learning Methods To Laser Acceleration Of Protons: Lessons Learned From Synthetic Data, Ronak Desai, Thomas Zhang, J. J. Felice, Ricky Oropeza, Joseph R. Smith, Alona Kryshchenko, Chris Orban, Michael L. Dexter, Anil K. Patnaik

Faculty Publications

In this study, we consider three different machine-learning methods—a three-hidden-layer neural network, support vector regression, and Gaussian process regression—and compare how well they can learn from a synthetic data set for proton acceleration in the Target Normal Sheath Acceleration regime. The synthetic data set was generated from a previously published theoretical model by Fuchs et al. 2005 that we modified. Once trained, these machine-learning methods can assist with efforts to maximize the peak proton energy, or with the more general problem of configuring the laser system to produce a proton energy spectrum with desired characteristics. In our study, we focus …

Towards A Comprehensive Metaverse Forensic Framework Based On Technology Task Fit Model, Amna Almutawa, Richard Adeyemi Ikuesan, Huwida Said

All Works

This article introduces a robust metaverse forensic framework designed to facilitate the investigation of cybercrime within the dynamic and complex digital metaverse. In response to the growing potential for nefarious activities in this technological landscape, the framework is meticulously developed and aligned with international standardization, ensuring a comprehensive, reliable, and flexible approach to forensic investigations. Comprising seven distinct phases, including a crucial incident pre-response phase, the framework offers a detailed step-by-step guide that can be readily applied to any virtualized platform. Unlike previous studies that have primarily adapted the existing digital forensic methodologies, this proposed framework fills a critical research …

Holistic Correlation Measure For Enhanced Encapsulation Of Trait Heterogeneity And Discovery Of Co-Expression, Zachary Valleroy

Theses

Large-scale, high-dimensional data analyses can be computationally prohibitive due to combinatorial explosion of the search space for finding complex patterns; a viable alternative is network modeling for abstraction and quantifying intrinsic data associations. Prominent network analysis methods furnish frameworks for model synthesis and validation but rely on standard correlation measures impaired by semi-supervised biases, latent heterogeneity, and uneven discretization techniques. Here we investigate a holistic measure for encapsulating data heterogeneity for enhanced efficacy of revealing complex patterns through network analysis. Our unique correlation metric, K-medoids Utility for Duo Original Similarities (Kudos), exhaustively factors real-valued analyte data to compute …

Vision-Language Integration For Enhanced Locomotion Mode Prediction, Ehsan Ahmadi

LSU Master's Theses

Wearable exoskeletons offer significant potential in enhancing human mobility in industrial environments. However, their adaptability to dynamic, task-intensive settings presents challenges, especially in accurately predicting locomotion modes such as ladder climbing, stair navigation, low-space movement, and obstacle navigation. This research proposes a multimodal framework that integrates visual data and speech commands to improve locomotion mode prediction in unpredictable environments. Multimodal data was collected using smart glasses, capturing both the user’s perspective (field-of-view, FOV) and voice during locomotion tasks. State-of-the-art models—CLIP, ImageBind, and GPT-4o—process these visual and linguistic inputs to predict locomotion activities. The models were evaluated in zero-shot and fine-tuned …

Medilink: A Secure Blockchain Framework For Multi-Institutional Healthcare, Jorge Castillo, Qian Chen

Informatics and Engineering Systems Faculty Publications

The use of Electronic Medical Records (EMRs) in the healthcare industry has proven to be critical for storing highly sensitive information. Disseminating and protecting healthcare data poses major challenges for the current healthcare information system. Blockchain technology provides solutions to these challenges with its inherited properties, such as decentralization, immutability, and transparency. This provides a unique opportunity to improve data sharing among stakeholders. We propose MediLink, a blockchain-based framework for secure collaborative medical storage. MediLink is designed to (1) protect EMR data from cyber attacks, (2) share healthcare information of patients with different stakeholders, and (3) enable the Internet of …

Dynamic Knowledge Elicitation: Leveraging Student Feedback For Improved Language Model Distillation, Reuven Muller

Master's Theses

Large Language Models (LLMs) have significantly advanced the field of natural language processing but remain resource-intensive and impractical for many organizations. Specialist models offer a viable alternative, often developed through Knowledge Distillation (KD) techniques. However, traditional KD methods rely on predefined static datasets to elicit knowledge from the teacher model, failing to dynamically address the weaknesses of the student model during training. This research introduces two novel methods for adaptive knowledge elicitation: Feedback-Driven Question Generation and Agent-Based Targeted Question Generation. These methods iteratively expand the training dataset based on the student model’s performance, leveraging a teacher model to generate targeted …

Gmr-216 Ai/Ml-Based Water Quality Monitoring Mobile App For Predicting E.Coli In Surface Waters, Keerthi Priya Karumanchi

C-Day Computing Showcase

E.coli contamination in surface waters has proven to be a significant public health concern, requiring innovative monitoring solutions. This paper presents the design of an AI-driven mobile application to predict whether E.coli bacteria are present at levels exceeding acceptable thresholds in surface waters. The methodology employs sensor devices to collect water quality data parameters, such as water temperature, pH, dissolved oxygen, and turbidity. A dataset is generated based on these parameters, and machine learning (ML) algorithms are applied to evaluate accuracy, precision, recall, and processing time. Additionally, our ML algorithms establish a correlation matrix among water quality parameters to identify …