IFIP Open Access Digital Library
Not a member yet
22614 research outputs found
Sort by
Compressed VGGNet for Automatic COVID-19 Disease Detection from CT Scan Images
No full textPart 1: SDG 3 Good Health and Well-BeingInternational audienceMillions of people across the globe were affected by this rapidly spreading disease by the year 2020. Contamination of the respiratory system results from contracting COVID-19. Nonetheless, establishing a conclusive diagnosis of COVID-19 may prove difficult due to the subtle differences it shares with conventional pneumonia and the complexities involved in identifying areas of infection. A growing number of approaches based on deep learning (DL) are being proposed for the automated detection of COVID-19 from CT scan images. By employing the data pretreatment methodology, unprocessed image data becomes prepared for further analysis. The proposed framework utilised transfer learning to construct VGGNet, which is capable of discerning the Covid-19 disease. The model was subsequently assessed in comparison to the most sophisticated models, VGG16 and VGG19, using the SARS-COV2 CT scan dataset. This dataset contains 1230 CT scans of non-infected images and 1252 CT scans of Covid-19-infected images. The model has achieved 99% accuracy, 98% precision, 97% recall, and 98% F1-score, among other performance metrics
DestiNexus: Probabilistic Framework for Weather- Informed Travel Itinerary Optimization
No full textPart 4: SDG 11 Sustainable Cities and CommunitiesInternational audienceDestiNexus is revolutionizing travel planning by utilizing a combination of advanced models to provide comprehensive and accurate recommendations. These include the cutting-edge RandomForestRegressor model for precise weather predictions, a RandomForestClassifier for accurate binary rain prediction, and an ARIMA model for effective time-series forecasting. The integration of ReactJS Front End and Backend Server allows for seamless collaboration and personalized suggestions based on factors such as location, date, and time. With a multi-model approach, DestiNexus not only offers specific predictions but also utilizes probabilistic modeling to provide users with valuable prediction intervals. Combining the power of machine learning, probabilistic techniques, and user-friendly design, DestiNexus offers a dynamic and immersive travel planning experience. Rigorous evaluation metrics, including Mean Squared Error and classification accuracy, underscore the reliability of its predictions. In essence, DestiNexus sets a new paradigm in travel planning, catering to a more nuanced and informed exploration of the world
Touchless Doorbell with Face Detection
No full textPart 4: SDG 11 Sustainable Cities and CommunitiesInternational audienceThe significance of infrastructure development within households is incomplete without the implementation of an efficient home security solution. Many existing security systems grapple with challenges such as lack of seamless integration, diminished accuracy, and a user-unfriendly experience, often compounded by high costs. Thus, the need for an innovative and effective approach to enhance the safety and security of homes becomes imperative. A groundbreaking solution to address these concerns is the integration of smart IoT-based touchless doorbells with face-detection capabilities. This research endeavors to propose a touchless doorbell system incorporating a face recognition model developed using OpenCV and a Haar cascade classifier. The integration of this system with an Arduino UNO further enhances its functionality. Through this innovative approach, a remarkable improvement in efficiency and accuracy, reaching approximately 93%, has been achieved. This, in turn, contributes to the establishment of a highly reliable and secure home security system. A noteworthy aspect of this research is its positive impact on the safety of visually impaired individuals. The touchless doorbell system has been augmented with features from the Google Text-to-Speech (gTTS) library, providing audio-related functionalities. This not only bolsters the overall accessibility of the security system but also ensures that visually impaired individuals can interact and navigate with the proposed home security system, thereby fostering inclusivity in the realm of home security solutions
Analysing Websites Privacy Policies: A Study of E-commerce Websites in South Africa
No full textPart 2: PrivacyInternational audienceWebsite privacy policies are used to inform consumers of the use and processing of their data but are often long and jargonised, complicating the comprehension of the website's privacy policies. While consumers require assurance on the processing of their personal information, organisations must also ensure that their website privacy policies cover the principles of data protection Acts. In South Africa, the Protection of Personal Information Act (POPIA) came into effect in July 2021, which provided principles for processing personal information. This study adopts the PRISMA methodology to systematically examine the existing literature to propose consolidated guidelines to aid website developers and administrators in drafting website privacy policies for the e-commerce sector. The proposed guidelines are a holistic consolidation of literature that applies to various jurisdictions. As the study was conducted in South Africa, the guidelines were also mapped to POPIA, and a sample of website privacy policies in South Africa were reviewed using the proposed guidelines. The e-commerce industry can benefit by implementing recommendations to aid them in addressing data privacy principles in website privacy policies
Gamification in Cybersecurity Training: High-Level Properties of Cybersecurity Games
No full textPart 1: Awareness and EducationInternational audienceThis paper examines high-level gamification properties, including mechanics, principles, engagement, and cybersecurity considerations suitable for educational settings. Utilising a literature review, the study consolidates these facets. Through this synthesis, the paper aims to present a unified understanding of gamification’s theoretical constructs and its pragmatic implications in education, specifically focusing on imparting cybersecurity concepts. A set of five properties that describe gamification in cybersecurity training is identified. The properties are described, and the relationship between the properties is described. The properties and their relationships form a foundation when developing cybersecurity training games
Security and Privacy Perspectives on Using ChatGPT at the Workplace: An Interview Study
No full textPart 2: PrivacyInternational audienceThe emergence of the artificial intelligence (AI) tool ChatGPT has created great excitement and unprecedented potential in various fields. Users are increasingly recognizing its benefits in aiding with work-related tasks and are incorporating it into their work routines. However, unconscious use of ChatGPT poses a risk to an organization if employees inadvertently disclose sensitive information. To date, there is a lack of research examining individuals’ perceptions of the security and privacy implications of ChatGPT use in organizational contexts. To bridge this gap, this study examines employees’ perceptions of security and privacy-related risks of using ChatGPT for work-related tasks and their strategies to mitigate these risks. Employing grounded theory, we conducted semi-structured interviews with 17 participants from 15 organizations across a range of professions and industries. Our findings indicate that employees have a general awareness of security and privacy-related risks, albeit with some uncertainties and misconceptions. While organizational guidelines for managing these risks are largely absent, participants describe that they employ self-determined strategies to avoid sharing sensitive data
Linux Kernel Keyloggers and Information Security
No full textPart 3: Technical Attacks and DefensesInternational audienceThis research paper aims to build and explore a Linux kernel module capable of logging keystrokes that a user would make on a Linux-based system. The module captures credentials which is a process known as keylogging. The kernel of the operating system manages all resources and data, and a breach in this area is a serious information security risk. This paper provides substantial evidence that kernel-level keyloggers are a very serious risk to information security in operating systems and computer systems in general. Such keyloggers can log user information, such as passwords, usernames and other information without much of the user’s knowledge
A Unified Privacy and Permission Management Framework
No full textPart 4: Usable SecurityInternational audienceThe increase in online services and digital channels has led to a large accumulation of user data, thus compromising data privacy. Researchers in the field of cybersecurity are seeking guidelines and solutions to protect user privacy as data processing by service providers becomes more extensive. Ensuring user privacy is the key to data protection, and providing users with the means to control their data remains the most effective method. To address the current complexity and proliferation of privacy settings, the authors developed the Unified Privacy and Permission Management Framework. This is a user-centric approach that simplifies the decision-making process and enhances the usability of privacy controls. It is built upon empirical insights and open-ended questions to understand users’ knowledge, perceptions and behaviours. The framework empowers users by presenting a streamlined, intuitive interface that facilitates informed decision-making and provides meta-level settings. This paper provides an example of how it can be applied in real-world scenarios to enhance user experience. The framework enables informed decision-making by providing a simple and intuitive interface that simplifies the complexity of privacy settings. Through this scenario, we illustrate the significant benefits users can experience, highlighting the framework’s potential to transform privacy management in the era of smart devices
Residential Price Analysis Using Machine Learning
No full textPart 2: Data AnalyticsInternational audienceThe main aim of the project is to gain insight into the decision-making process of housing prices using real estate data and machine learning techniques. This requires data analysis to identify key trends and trends regarding changes in house prices. The main goal is to create a reliable prediction model capable of predicting the future price of the house. This model will provide valuable advice to home buyers and sellers, enabling them to make informed decisions regarding real estate transactions. Finally, the successful program will enable individuals, real estate professionals and investors to tap into the complex real estate market. The project will help make more informed decisions by understanding the value of the house, develop investment strategies and contribute to business transparency. Analysing real estate prices is a difficult task because there are many factors that affect it. Factors such as location, property type, amenities and price are important to buyers. Accurate estimates are important to help people find housing within their budget without compromising their financial security. Machine learning algorithms can help you make informed decisions when choosing a home. Comparing regression methods such as linear, random forest, XGBoost and other regression, this model aims to predict high population housing price. The main success of this model is to accurately predict the price of the house based on the customer’s needs. In this study, various machine learning algorithms are tried to be used to predict real estate prices. The algorithm with the most accurate prediction will be selected for use
Deep Neural Network Based Relocalization of Mobile Robot in Visual SLAM
No full textPart 4: Intelligent RobotInternational audienceRelocalization after track loss in Visual Simultaneous Localization and Mapping (Visual SLAM) is a critical challenge in robotics, especially for autonomous mobile robots navigating dynamic environments. This paper introduces a deep learning approach that employs Deep Neural Networks (DNNs), particularly VGG16 and ResNet34, to reorient and relocalize robots effectively. Trained on a vast repository of indoor images from the Multimodal Indoor Simulator (MINOS) and Matterport3D dataset, the DNN models discern the most viable direction for movement be it translation or rotation based on the robot’s current visual input in relation to its last known position within the ORB-SLAM2 generated map. The methodology involves real-time data exchange between MINOS and the ORB-SLAM2 system via a dedicated ROS node, facilitating the recovery process. Extensive testing shows that our proposed model successfully predicts the appropriate recovery action in over 90% of track loss instances, substantiating its efficacy and potential for deployment in real-world applications. This research contributes to the advancement of robust relocalization strategies in Visual SLAM, enhancing the autonomous capabilities of mobile robotics