1393 research outputs found
Sort by
A Comprehensive Framework Empowering Organizations to Effectively Minimize the Risk of Big Data De-anonymization Attack
Get PDFIn the era of big data, ensuring the privacy and security of sensitive information has become a paramount concern for organizations across various sectors. One of the significant threats to data privacy is the risk of de-anonymization attacks, which can compromise the anonymity of individuals within large datasets. In response to this challenge, this dissertation proposes a comprehensive framework designed to empower organizations to effectively minimize the risk of big data de-anonymization attacks. The core objective of this research is to develop an anonymization techniques assessment framework that evaluates the vulnerabilities associated with anonymization techniques commonly used by organizations. By assessing these vulnerabilities, the framework aims to provide organizations with actionable insights to enhance their data anonymization practices and protect against de-anonymization attacks.
The proposed framework leverages the expertise of subject matter experts in the field of data privacy and security. Through a rigorous evaluation process, these experts validate the effectiveness of the framework as a practical tool for organizations seeking to safeguard their data against de-anonymization attacks. The development of this framework represents a significant contribution to data privacy and security. By providing organizations with a systematic approach to assess and mitigate the risks of de-anonymization, the framework has the potential to enhance data protection measures and promote trust in the use of big data analytics.
Overall, this dissertation aims to bridge the gap between theoretical knowledge and practical application in the realm of big data de-anonymization prevention. This research strives to pave the way for a more secure and privacy-conscious data ecosystem by equipping organizations with the tools and insights needed to safeguard their data privacy
Understanding Provider Perspectives: Catalysts and Challenges of Intelligent Clinical Decision Support Systems
Get PDFThe study aims to understand the providers\u27 perspectives of Intelligent Clinical Decision Support Systems (ICDSS). Providers are the key users of ICDSS. ICDSS uptake is heavily reliant on providers. However, providers have a powerful sense of personal identity due to their prestige, high autonomy, and socially accepted role. Currently, providers are resisting ICDSS. ICDSS resistance is due to poor fit with workflow, lack of relevant output and usefulness, insensitivity to complex patients, autonomy challenges, and an impact on the provider-patient relationship. Recently, researchers have overlooked the providers\u27 contextual perspectives of ICDSS. The study addressed the research gap by understanding the providers\u27 perspectives on their experiences, the ICDSS clinical utility, and the providers\u27 adaptation to the ICDSS. No known research exists explaining a comprehensive understanding of the providers\u27 perspectives of ICDSS. Therefore, this research study empirically investigated the providers\u27 perspectives to lay the foundation for tailoring ICDSS to providers\u27 unique needs and preferences. As a result, provider satisfaction and patient care will improve. The study used Grounded Theory as an overarching methodology for our qualitative case study research. Semi-structured interviews were used to gather the providers\u27 narrative data. The researcher transcribed the interview data and further refined it by comparing the text data to the audio recording data. The researcher analyzed the transcribed data using open coding. Further analysis identified catalysts and challenges for ICDSS. The ICDSS catalysts facilitated medical search, improved task efficiency, leveraged data-driven intelligence, provided transparent and reliable recommendations, improved treatment choices, prevented adverse events, supported their diagnosis process, promoted personalized training, promoted team dialogue, and improved technical literacy. The challenges of the ICDSS included contributing to cognitive overload, data quality issues, inflexibility, dehumanization of the patient-provider interaction, diminishing provider autonomy, increasing workload through compliance requirements, misalignment with their workflow, increasing AI anxiety, challenging to learn, and a resistance to change. The researcher grouped the focused codes into provider experiences, clinical utility, and adaptation. These were the categories of the ICDSS-Diagnostic Perspective Model (ICDSS-DPM). The ICDSS-DPM proposed six propositions that indicated the providers\u27 perspectives of ICDSS
Weed Detection Using Lightweight Deep Learning Models with Transfer Learning and Hyperparameter Optimization
Get PDFIn agriculture, the presence of weeds adversely impacts yield. Confronting this challenge manually requires continuous and labor-intensive field monitoring processes. Artificial intelligence, notably deep learning is one the most promising for weed detection, particularly when deployed to edge devices. However, these edge devices have limited computational resources, making it challenging to deploy large Deep Learning (DL) models often resulting in a performance-size trade-off. This research proposes a three-pronged approach that leverages DL lightweight architectures (LWA) to optimize for size and minimize computational resources at the edge, transfer learning to mitigate constraints associated with limited data commonly encountered in this domain, and Bayesian optimization for hyper-parameter tuning.
Evaluation of the proposed approach reveals that, compared to past research, the lightweight architectures utilized in this study surpassed the previously best-performing state-of-the-art ResNet152V2 model. Notably, the LWA EfficientNetV2B0 model achieved an accuracy of 95.79% with a model size of 23 MB, while LWA MobileNetV2 attained an accuracy of 95.09% with a model size of 13.96 MB. Overall, LWA MobileNetV2 attained an accuracy nearly on par with EfficientNetV2B0, while utilizing a substantially smaller model. Such efficiency is crucial for edge devices constrained by their computational capacity.
Implementing advanced deep learning models within weed detection systems can incur substantial costs. The findings from this research contribute towards enhancing both performance and accessibility. The proposed approach demonstrates the efficacy of LWA coupled with transfer learning and hyper-parameter tuning to address the unique needs of the domain while producing better accuracy results and considerable reduction in model sizes compared. The approach can be generalized to other problem domains with similar constraints. In regard to practice, the proposed approach provides a technological solution that is more affordable and accessible to mid and low-tier farmers, thereby enabling them to increase productivity by effectively addressing weed infestation during the critical early stages of crop growth
Generative Adversarial Networks in Fraud Detection: A Systematic Literature Review
No full textFraud detection has become an important part of modern society. Traditional fraud detection techniques struggle with unbalanced data because of the lack of data representing fraudulent behaviors. Recently, Generative Adversarial Networks (GANs) have attracted much attention in producing synthetic data to balance a dataset. In this paper, we present a systematic review of the literature in the area of GAN applications in fraud detection. This paper analyzes the relationships between fraud detection and GANs, and identifies the roles of GAN usage regarding fraud detection: feature-based (85%) and image-based (15%). In addition, the most used GAN architecture is the standard vanilla GAN (37.5%), and the most common fraud aspects for GAN applications are credit card fraud (42.5%), financial fraud (15%), and identity fraud (15%)
Critical Success Factors for BI Systems Implementation and Delivery: A Systematic Literature Review
Get PDFDespite the widespread acquisition of business intelligence (BI) systems, their implementation has not always been successful. This study examines the critical success factors (CSFs) that impact the implementation of BI systems in organizations. The systematic literature review follows the guidelines of Kitchenham and Charter’s research that was published in 2007. A total of 93 articles published between 2011 and 2021 were analyzed for CSF related to BI systems implementation and delivery. The study identified 56 CSFs linked to organization empowerment & operations, 52 CSFs related to system implementation, and 28 CSFs associated with user enablement. The study found a paucity of research on user enablement in the context of BI implementation and delivery, highlighting a gap in the literature. The findings of this study can help organizations better understand the factors that contribute to successful BI system implementation and delivery, and guide future research in this field
Health Benefits and Adverse Effects of Kratom: A Social Media Text-Mining Approach
Get PDFBackground: Kratom is a substance that alters one’s mental state and is used for pain relief, mood enhancement, and opioid withdrawal, despite potential health risks. In this study, we aim to analyze the social media discourse about kratom to provide more insights about kratom’s benefits and adverse effects. Also, we aim to demonstrate how algorithmic machine learning approaches, qualitative methods, and data visualization techniques can complement each other to discern diverse reactions to kratom’s effects, thereby complementing traditional quantitative and qualitative methods. Methods: Social media data were analyzed using the latent Dirichlet allocation (LDA) algorithm, PyLDAVis, and t-distributed stochastic neighbor embedding (t-SNE) technique to identify kratom’s benefits and adverse effects. Results: The analysis showed that kratom aids in addiction recovery and managing opiate withdrawal, alleviates anxiety, depression, and chronic pain, enhances mood, energy, and overall mental well-being, and improves quality of life. Conversely, it may induce nausea, upset stomach, and constipation, elevate heart risks, affect respiratory function, and threaten liver health. Additional reported side effects include brain damage, weight loss, seizures, dry mouth, itchiness, and impacts on sexual function. Conclusion: This combined approach underscores its effectiveness in providing a comprehensive understanding of diverse reactions to kratom, complementing traditional research methodologies used to study kratom
A Comparative Analysis of the Interpretability of LDA and LLM for Topic Modeling: The Case of Healthcare Apps
Get PDFThis study compares the interpretability of the topics resulting from three topic modeling techniques, namely, LDA, BERTopic, and RoBERTa. Using a case study of three healthcare apps (MyChart, Replika, and Teladoc), we collected 39,999, 52,255, and 27,462 reviews from each app, respectively. Topics were generated for each app using the three topic models and labels were assigned to the resulting topics. Comparative qualitative analysis showed that BERTopic, RoBERTa, and LDA have relatively similar performance in terms of the final list of resulting topics concerning human interpretability. The LDA topic model achieved the highest rate of assigning labels to topics, but the labeling process was very challenging compared to BERTopic and RoBERTa, where the process was much easier and faster given the fewer numbers of focused words in each topic. BERTopic and RoBERTa generated more cohesive topics compared to the topics generated by LDA
Exploring Machine Learning with FNNs for Identifying Modified DGAs through Noise and Linear Recursive Sequences (LRS)
No full textThe study proposes a comprehensive technique to identify novel variations within Domain Generating Algorithm (DGA) families, crucial for securing critical infrastructures. This technique incorporates Damerau-Levenshtein Distance to enhance Feedforward Neural Network (FNN) adaptability to diverse DGA manifestations, including Linear Recursive Sequence (LRS) modification. By strategically selecting features, it demonstrates robustness against domain-specific noise, vital in detecting increasingly sophisticated cyber threats. The approach is systematically evaluated for adaptability to various noise forms, ensuring real-time threat detection and incident response efficacy. With an accuracy rate of 100%, the method proves its versatility in handling diverse cyber threats, making it a valuable asset for network security practitioners
Forensic Iconography for Image Forgery Detection
Get PDFTraditionally, image forgeries were limited to defraud art collectors and falsify documents. The rise of visual media provided legitimate reasons for artists to forge images in the pursuit of more effective storytelling. In all cases these methods required specialized domain knowledge and highly skilled artists. The rise of computers simplified many of these processes, lowering the bar in time, materials, and skill, and thus giving most users access basic tools for altering images. Today, image alteration tools leverage machine learning and artificial intelligence techniques, and can generate incredibly accurate image forgeries. Effective methods for verifying the authenticity of media data are increasingly needed. This work seeks to identify and explain methods for identifying video image forgeries, more commonly known as “deepfakes.
An Algorithm Based on Priority Rules for Solving a Multi-drone Routing Problem in Hazardous Waste Collection
Get PDFThis research investigates the problem of assigning pre-scheduled trips to multiple drones to collect hazardous waste from different sites in the minimum time. Each drone is subject to essential restrictions: maximum flying capacity and recharge operation. The goal is to assign the trips to the drones so that the waste is collected in the minimum time. This is done if the total flying time is equally distributed among the drones. An algorithm was developed to solve the problem. The algorithm is based on two main ideas: sort the trips according to a given priority rule and assign the current trip to the first available drone. Three different priority rules have been tested: Shortest Flying Time, Longest Flying Time, and Median Flying Time. Two recharging conditions are maintained: recharging needed time and recharging full duration. By applying each priority rule and each recharging condition, we generate a six versions of the algorithm. The six versions of the proposed algorithm were implemented in Java programming language.The results were analyzed and compared proving that the Longest Flying Time priority rule surpasses the other two rules. Moreover, recharging a drone just enough for taking the next trip proved to be better than fully recharging it