IFIP Open Access Digital Library
Not a member yet
22614 research outputs found
Sort by
HyPRETo: Hybrid Pre-trained Ontology Approach for Contextual Relation Classification on Mosquito Vector Biocontrol Agents
No full textPart 3: SDG 9 Industry, Innovation and InfrastructureInternational audiencePre-trained Language Model facilitates contextual relation classification by capturing contextual information, addressing word ambiguity, encoding global sentence context, enabling transfer learning, handling out-of-vocabulary words, and improving performance with limited labelled data. Existing pre-training approaches suffer in size, bias, interpretability, generalization, and the lack of domain specificity. To address this, HyPRETo, the hybrid model that combines the strength of token replacement and dynamic masking is proposed to achieve upgraded performance to increase classification accuracy. The Mosquito Vector Biocontrol Agents data is used for implementing the model for a contextual relation classification task. HyPRETo uses ontology to provide structured knowledge. HyPRETo is pre-trained by ELECTRA and fine-tuned by RoBERTa models. Feedforward and softmax activation function is used for classification. The Natural Language Processing technique and SQL database are used to develop an automated question-answering system. The HyPRETo was evaluated with state-of-art models and achieved 98.42% accuracy. As a contribution, the manually annotated input dataset on the mosquito vector control agent is prepared for the classification task. Subsequently, the enhanced model is developed. The interface for an automated question-answering system for mosquito vector biocontrol agents is developed to assist public health applications such as mosquito vector control, disease control, ecosystem management, environmental conservation, and so on
A Deep Learning-Based Algorithm for Predicting the Turning Point of Cloud Workload
No full textPart 3: SDG 9 Industry, Innovation and InfrastructureInternational audienceCloud workload data is time series data that consists of the logs of the resource consumption, such as central processing unit (CPU) utilization, memory utilization and bandwidth consumption, on a particular machine in a cloud environment. These logs are essential for analyzing the machine’s performance and debugging purposes at fault time. It is worth noting that the utilization fluctuates due to changes in load over time. Moreover, there are some peaks and valleys in the utilization pattern, and predicting such points in the cloud workload is challenging and not well-studied. The prediction can be helpful in saving the waste of resources and providing quality services to the customers. Recently, researchers have used deep learning algorithms to predict the turning points using basic and fluctuating features. However, the prediction can be improved by taking the extreme points of the adjacent segments, and it is effective for high-variance time series data. This paper presents a turning point prediction (TPP) algorithm on cloud workload data using deep learning models, namely long short-term memory (LSTM), bidirectional LSTM (BiLSTM) and gated recurrent unit (GRU) networks, that use some novel features. TPP uses a piece-wise linear segmentation (PLS) algorithm to generate the segments. Then, it determines the potential turning points and labels them into two classes. Finally, the LSTM, BiLSTM, and GRU are applied to classify the turning points. We simulated the TPP on two datasets, Google cloud and Alibaba, and compared its three variants in terms of precision, recall and F1-score to show its effectiveness
Enhancing Adaptive Data Management Middleware: Techniques for Federated Cloud Environments
No full textPart 3: SDG 9 Industry, Innovation and InfrastructureInternational audienceThe adaptive data management middleware is crucial for facilitating efficient and dynamic data management in distributed computing environments. Nevertheless, the widespread adoption of federated cloud environments brings about new complexities and difficulties in data administration, requiring inventive methods to improve the functionalities of current middleware systems. This research paper introduces innovative strategies for enhancing adaptive data management middleware that is especially designed for federated cloud environments. The proposed methodology addresses several fundamental components of adaptive data management middleware. Initially, proposed the methods for dynamically provisioning and allocating resources to enhance the efficiency of data processing and storage across federated cloud environments. In addition, provided the strategies for implementing data governance and enforcing security measures to meet regulatory standards and protect confidential information in federated cloud environments. Furthermore, investigated the techniques for duplicating and synchronization data to improve the accessibility and uniformity of data in federated environments. To evaluate the results of the proposed techniques, we conducted comprehensive experimental evaluations using simulated federated cloud infrastructures. The results demonstrate the substantial enhancements in the performance of data management, utilisation of resources, and capacity to scale when compared to existing methods. Lastly discussed the implications of our findings and emphasised potential areas for further investigation in this rapidly evolving domain
Cluster Based Classification of Question Independent C Codes
No full textPart 3: SDG 9 Industry, Innovation and InfrastructureInternational audienceIn the field of software development, ensuring the accuracy and quality of code remains a paramount concern. The task of precisely classifying code as correct or incorrect poses inherent challenges. This research introduces a groundbreaking approach that uses clusters constructed from code embeddings generated by CodeBERT to effectively classify code into distinct clusters representing correctness or incorrectness. The model’s ability to grasp intricacies in code semantics and structure leads to a significant reduction in debugging efforts. Consequently, this approach contributes to an overall increase in the reliability and robustness of software systems. CATBOOST algorithm consistently demonstrated high performance with an accuracy of 84% during the experimentation
Ethereum Powered Web3 Crowdfunding Platform
No full textPart 3: SDG 9 Industry, Innovation and InfrastructureInternational audienceA project or endeavor can raise money by using a technique called crowdfunding, which involves asking a large number of people to make small donations. Centralized applications, expensive taxes, a lack of transparency, and security vulnerabilities are the platform’s current drawbacks. Blockchain is utilized to get around these constraints. This distributed database serves as a public ledger, of all completed transactions and digital events that are shared amongst involved parties. Smart contract technology is another benefit of blockchain. This platform allows users to generate funds for projects or other purposes by providing proper documentation supporting their identity. By confirming that the documents are legitimate for voting and funding, authorized individuals can donate money to the various initiatives they support. Smart contracts will forward monies gathered to the appropriate fundraisers if the majority of donors approve. When their limit is reached, they will be able to access their money using a smart contract. The blockchain keeps track of every transaction. The platform will be safer and more transparent with the use of these smart contracts. Better than the conventional apps, this offers consumers access to a fully functional and safe crowdfunding online application platform
Data After Death: Australian User Preferences and Future Solutions to Protect Posthumous User Data
No full textPart 2: PrivacyInternational audienceThe digital footprints of today’s internet-active individuals are a testament to their lives, and have the potential become digital legacies once they pass on. Future descendants of those alive today will greatly appreciate the unprecedented insight into the lives of their long-since deceased ancestors, but this can only occur if today we have a process for data preservation and handover after death. Many prominent online platforms offer nebulous or altogether absent policies regarding posthumous data handling, and despite recent advances it is currently unclear who the average Australian would like their data to be managed after their death (i.e., social media platforms, a trusted individual, or another digital executor). While at present the management of deceased accounts is largely performed by the platform (e.g., Facebook), it is conceivable that many Australians may not trust such platforms to do so with integrity. This study aims to further the academic conversation around posthumous data by delving deeper into the preferences of the Australian Public regarding the management of their data after death, ultimately to inform future development of research programs and industry solutions. A survey of 1020 Australians revealed that most desired a level of control over how their data is managed after death. Australians currently prefer to entrust the management of their data to a trusted close individual or a third-party software that they can administrate themselves. As expected, social media companies ranked low regarding both trust and convenience to manage data after death. Furthermore, we found that the more active internet users have stronger desire for control over their data after death, as did people with children and those with greater levels of formal education. Unexpectedly, marital status, age, and gender did not predict preferences for posthumous data control. Future research focus should be to conceptualise and develop a third-party solution that enables these preferences to be realised. Such a solution could interface with the major online vendors (social media, cloud hosting etc.) to action the deceased’s will – erasing select data, while sharing other data with selected individuals
Intrusion Detection System Trends: An Overview of Current Advances in IoV & Communication Networks
No full textPart 1: Applications of AI/ML in KDM, Cloud Computing & SecurityInternational audienceThe proliferation of Internet of Vehicles (IoV) indeed has paved the way for an inter-connected system of transportation, offering unprecedented convenience and efficiency. However, this network is susceptible to various intrusions of malicious means. The volatility of security in IoV arises from the interconnected and dynamic nature of vehicular communication. With the increasing number of connected vehicles and the continuous evolution of communication protocols, the attack surface expands, rendering traditional security measures inadequate. The potential consequences of security breaches in IoV are far-reaching, encompassing safety risks, privacy infringements, and disruptions to critical transportation infrastructure. Threat identification systems emerge as a critical defense mechanism, along with firewalls and other security measures. This paper assesses a comprehensive overview of existing IDS approaches to cater to different kinds of intrusions, implemented in IoV environments or in a different setting. The study provides a comprehensive overview of existing IDS approaches, evaluating their effectiveness in IoV settings as well as in alternative contexts. The discussion encompasses various aspects of intrusion detection within the IoV framework
Analyzing Meteorological Data for Curated Music Selection
No full textPart 2: Data AnalyticsInternational audienceAnalyzing Meteorological Data for Curated Music Selection, delves into the innovative fusion of weather data and music curation. Leveraging advanced data analysis and machine learning techniques, we aim to create personalized playlists that align with the current weather conditions. Our research addresses the challenge of correlating weather elements with musical attributes and explores real-time adaptation, potentially revolutionizing the way individuals experience and interact with music in various meteorological contexts. This study not only advances the field of data-driven music curation but also provides insights into its transformative impact on music services and user experiences
Review on Biomedical Informatics Through the Versatility of Generative Adversarial Networks
No full textPart 3: Applications of MLInternational audienceBiomedical informatics has recently seen the rise of generative adversarial networks (GANs), which provides a versatile approach to addressing diverse challenges and propelling advancements in personalized medicine. This review paper explores the transformative potential of GANs, delving into their applications across various domains of biomedical research and clinical practice. GANs, characterized by two competing neural networks—the generator and the discriminator—operate in an adversarial setting. The generator strives to produce realistic data samples from a specified distribution, when the discriminator attempts to differentiate between authentic and false input. This adversarial setup compels the generator to continually enhance its ability to generate realistic data. The versatility of GANs arises from their capacity to learn intricate data distributions and generate authentic synthetic data. The multidisciplinary area of biomedical informatics combines information technology, computer science, and life sciences to handle and evaluate medical data. This dataset spans multiple domains. In biomedical informatics, GANs find applications in medical image analysis, enhancing images for improved segmentation, noise reduction, and increased resolution, leading to more accurate diagnoses and treatment planning. Additionally, The AI world has been enthralled with Generative Adversarial Networks (GANs) due to their capacity to provide varied and realistic data. This paper delves into the rapidly developing subject of GAN applications in biomedical informatics, emphasizing how they can revolutionize a number of fields. Moreover, GANs contribute to biomarker discovery from large biomedical datasets, aiding in early disease detection and diagnosis, thereby paving the way for personalized medicine. They also enable personalized treatment plans by predicting patient outcomes and tailoring therapies based on individual patient characteristics, optimizing healthcare interventions. Despite their promise, GANs face challenges such as interpretability, data privacy, and computational complexity, necessitating ongoing research efforts. Nonetheless, the immense potential of GANs to revolutionize biomedical informatics and advance personalized medicine remains evident. With continued GAN research, further breakthroughs are anticipated, promising transformative impacts on healthcare
DRL-SLAM: Enhanced Object Detection Fusion with Improved YOLOv8
No full textPart 5: Perceptual IntelligenceInternational audienceSimultaneous Localization and Mapping (SLAM) technologies are pivotal in advancing robotics and autonomous navigation, particularly within challenging indoor environments. These systems face significant challenges due to dynamic variables such as fluctuating lighting, occlusions, and the presence of moving objects. In response, this paper introduces a novel integration of Deep Reinforcement Learning (DRL) with the advanced object detection capabilities of YOLOv8 within a SLAM framework, termed DRL-SLAM YOLOv8. This integration enhances object detection by leveraging DRL’s ability to learn from environmental interactions and YOLOv8’s precise and efficient real-time object recognition. Our experiments demonstrate the superiority of DRL-SLAM YOLOv8 over traditional methods, with marked improvements in detection accuracy and system reliability under diverse conditions. Notably, our approach significantly improves navigational effectiveness, as evidenced by increased distance coverage and goal achievement compared to standard SLAM techniques, validating its potential in real-world applications