IFIP Open Access Digital Library
Not a member yet
22614 research outputs found
Sort by
An Interdisciplinary Thematic Analysis of the US National Guard Bureau Response to the SolarWinds Attack
No full textPart 3: Technical Attacks and DefensesInternational audienceThe SolarWinds attack of 2020 was one of the most impactful cyberattacks on the US. Our interdisciplinary research team had the opportunity to observe and analyze the human aspects of the corresponding incident response as it unfolded. Four main themes were identified through a series of interviews and incident observations. This led to an understanding of the importance of establishing the following for highly effective and efficient incident response teams: 1) a portfolio of tools for increasing communication, collaboration, comfort level, and cohesion, 2) a team with diverse education, training, and experience, especially military leadership experience, and 3) teams with long established relationships to achieve high levels of trust, cohesion, and resilience. Ultimately, this analysis resulted in recommendations for further enhancing teams operating at this scale and intensity
“There is a Damn Hello on the Social Media!” Insights into Public Risk Communication in the Digital Age
No full textPart 1: Management and RiskInternational audienceThe ability to handle threats, such as disinformation, manipulation of public opinion, and disruption of critical supplies, is becoming increasingly important, thus, necessitating, among other strategies, efforts to establish a proper risk communication to the public. This paper addresses the need for more empirical research in this area to contribute to the development of an in-depth understanding of public risk communication that includes information-related threats and cyber issues. The study involves officials of three public organizations entrusted with safety and security in society: the police, the rescue service, and the county administrative board of a county in the middle of Sweden. The results detail the recognition of risks to be communicated, the organization of the communication process, the messages that these actors seek to bring forth, and to whom as well as challenges of public risk communication in the digital era. The findings indicate that information-related and cyber risks are increasingly essential to consider as an additional layer of public communication. Two implications emerged as particularly important: (1) all communication about risks and crises must consider the systemic risk of mis- and disinformation, and (2) tailored communication about the risks interrelated with disinformation should use human-centered, dialogue-based, and moderated approaches. Further research can focus on associated challenges, considering the distribution of responsibilities, inter-organizational information sharing and cooperation, and the possibly stochastic effects on critical (information) infrastructures and, ultimately, societal values
Literature Review: Misconceptions About Phishing
No full textPart 2: Social EngineeringInternational audiencePhishing is a danger to both private users and businesses. Industry and academia have proposed several approaches to deal with this threat, many of which developed with a supposedly human-centric design. Yet, to our knowledge, there is no research focused on the misconceptions that users might have on phishing. This glaring gap is a problem, as previous research has shown that not engaging with the mental model of users can lead to lack of effectiveness of an approach in the real world. To address this gap, we conducted a systematic literature review starting from papers published at CHI in the last ten years, and expanding to other venues through a backward and a forward search based on the initial relevant CHI papers. We identified 15 misconceptions about phishing in 21 papers that researchers should address in their solutions to enhance the effectiveness of their approaches
Computational Intelligence in Data Science: 7th IFIP TC 12 International Conference, ICCIDS 2024, Chennai, India, February 21–23, 2024, Revised Selected Papers, Part II
No full textInternational audienceBook Front Matter of AICT 71
Exploring Corrosion Detection: Deep Learning and Ensemble Approaches Analysis
No full textPart 2: Data AnalyticsInternational audienceCorrosion is a widespread challenge in industrial and urban settings, silently causing damage to buildings, pipes, and machinery, leading to significant financial losses, safety risks, product quality issues, and operational inefficiencies. The goal of this research is to improve safety, quality, and efficiency in various areas by going beyond simple data analysis and statistics. It assesses different deep learning techniques and shows that the Conventional Convolutional Neural Network (CNN) is the best for corrosion detection. In order to further improve precision, the study combines a Random Forest classifier with a Conventional CNN model, leading to notable gains in accuracy and F1 score. The study also investigates the usage of Extra Trees with Conventional CNN, demonstrating the benefit of Bootstrap Aggregation in improving the accuracy of corrosion detection. These findings strengthen endeavors aimed at preserving infrastructure by significantly aiding in the detection of corrosion, with potential applications in drone technology and web application development for early corrosion detection
Evolving Financial Markets: The Impact and Efficiency of AI-Driven Trading Strategies
No full textPart 6: AI for ScienceInternational audienceThis paper investigates the impact of Artificial Intelligence (AI) on trading strategies in financial markets, comparing AI-driven approaches with traditional methodologies and their effects on market efficiency, liquidity, and volatility. It critically examines how AI challenges established financial theories, such as the Efficient Market Hypothesis and behavioral finance, suggesting a potential redefinition of market dynamics considering AI’s superior data processing and analytical capabilities. The study synthesizes academic literature, theoretical insights, and speculative analysis to assess the multifaceted implications of AI’s integration into trading practices. Findings highlight AI-driven strategies’ enhanced risk-adjusted returns and contribution to market efficiency, yet underscore a complex impact on market dynamics, where AI can both improve liquidity and introduce volatility. Ethical considerations and regulatory challenges are emphasized, pointing to the need for transparent and adaptive regulatory frameworks to address the opacity of AI decision-making and ensure market integrity. The paper advocates for interdisciplinary research and collaboration among technologists, regulators, and market participants to navigate the evolving landscape of AI in trading. Through this exploration, the study contributes to the discourse on AI-driven trading, balancing the benefits of technological advancements against the risks and challenges, and underscores the critical role of regulatory oversight in shaping the future of financial markets
Trajectory Prediction of Unmanned Surface Vehicle Based on Improved Transformer
No full textPart 4: Intelligent RobotInternational audienceIn recent years, the significance of applying Unmanned Surface Vehicles (USVs) in coastal defense has been progressively increasing. Precise prediction of USV trajectories plays a vital role in the decision-making of coastal defense, anti-privacy, and so on. However, the intricate nature of USV trajectories, characterized by high maneuverability and sudden motion pattern changes, poses great challenges for accurate prediction. To address these issues, this paper proposes a trajectory prediction model based on an improved Transformer with sparse self-attention and physical rule constraints. Focus on designing the “Max-Mean” sparse self-attention mechanism to streamline computational demands and memory usage, and the physical loss function to improve the accuracy and robustness of predictions. Moreover, a generative decoder is included to improve the model’s ability to process long sequence data and the inference efficiency. To verify the prediction effect of the proposed method, we construct a USV simulation trajectory dataset based on the ship kinematic model for trajectory prediction experiments. The simulation results illustrate that the proposed model surpasses existing trajectory prediction models and fulfills the stringent requirements for precise and rapid USV trajectory predictions
Supervised Learning of Procedures from Tutorial Videos
No full textPart 2: Applications of AI/ML in Image ProcessingInternational audienceOnline educational platforms and MOOCs (Massive Open OnlineCourses) have made it so that learning can happen from anywhere across theglobe. While this is extremely beneficial, videos are not accessible by everyone,due to time constraints and lower bandwidths. Textual step-by-step instructionswill serve as a good alternative, being less time-consuming to follow than videos,and also requiring lesser bandwidth. In thiswork, the authors aim to create a systemthat extracts and presents a step-by-step tutorial from a tutorial video, whichmakesit much easier to follow as per the user’s convenience. This can be mainly accomplishedby using Text Recognition for academic videos and Action Recognitionfor exercise videos. Text Recognition is accomplished using the Pytesseract packagefrom Python which performs OCR. Action Recognition is performed withthe help of a pre-trained OpenPose model. Additional information is extractedfrom both exercise and academic videos with the help of speech recognition. Theinformation extracted from the videos are then segmented into procedural instructionswhich is easy to comprehend. For exercise videos, a frame-wise accuracy of79.27% is obtained. Additionally, for academic videos, an accuracy of 78.47% isobtained
Lecter - A Large Language Model Chatbot for Cognitive Behavioral Therapy
No full textPart 1: Applications of AI/ML in Natural Language ProcessingInternational audienceMental Illnesses such as Anxiety and Depression are fairly prevalent in modern society. Yet most people do not get diagnosed, cannot afford treatment, or there simply aren’t any readily available practitioners. In the worst-case scenario, avoiding treatment can increase the likelihood of self-harm in affected individuals, both emotionally and physically. A low-cost and easily available solution is required to provide the benefits of therapy to the masses. In this paper, we introduce “Lecter”, a chatbot solution specifically targeted to providing free mental health care to individuals who cannot access an actual therapist. With the help of the superior conversational abilities of an LLM (large language model), Lecter listens to the concerns and problems that patients face and helps them introspect, and better understand their feelings using a popular therapy technique called Cognitive Behavioral Therapy (CBT). Lecter provides its services using technologies such as an LLM called Mythomax, React Js for the front end of the application, Flask for the back end, and MongoDB for authentication and storage
De-noising of Low Dose CT Liver Images Using Improved Discrete Wavelet Transform
No full textPart 2: Applications of AI/ML in Image ProcessingInternational audienceCT scans have become increasingly common in recent years and are essential for diagnosing various medical conditions. But the most significant element influencing a CT image’s quality is noise, which frequently taints low-dose scans. The radiologists’ choice may then be impacted by this. Therefore, in order to enhance visibility and increase the clarity of the image, LDCT images must be removed of noise before being utilized for diagnosis. This work is one such attempt to remove noise from low dose Computed Tomography (CT) liver images. In the present work, the noise from the CT liver images is eliminated using a modified discrete wavelet transform (DWT). The scheme system is used in conjunction with the DWT. The lifting approach improves the wavelet transform’s accuracy, simplicity, efficiency, and flexibility. Furthermore, the noisy wavelet coefficients are denoised via Bayesian shrinkage. The de-noised image PSNR, MSE, SSIM, and SNR values are computed, and a comparison with the traditional de-noising methods is carried out. The quantitative study reveals that the suggested approach has produced noteworthy outcomes when compared to the efficacy of conventional procedures. Furthermore, as the results and discussion section demonstrate, the suggested lifting technique in conjunction with the DWT minimizes the memory needs while simultaneously reducing complexity