International Journal of Communication Networks and Information Security (IJCNIS)
Not a member yet
1021 research outputs found
Sort by
CONVOLUTION NEURAL NETWORK-BASED SPEECH EMOTION RECOGNITION USING MFCCS
No full textSpeech Emotion Recognition (SER) plays a crucial role in applications related to affective computing and human-computer interaction. Historically, many techniques for emotion recognition relied on simple feature extraction paired with basic classifiers. However, these traditional methods often demonstrated limited effectiveness in accurately identifying emotions. To address these shortcomings, this paper proposes five distinct models based on Convolutional Neural Networks (CNN) for recognizing emotions from speech signals.
The methodology outlined in this approach focuses on recognizing seven emotions: disgust, neutrality, fear, joy, anger, sadness, and surprise. CNN is utilized alongside advanced feature extraction techniques, such as Pitch and Energy, Mel-Frequency Cepstral Coefficients (MFCC), and Mel Energy Spectrum Dynamic Coefficients (MEDC), to enhance recognition performance. These feature extraction methods have been shown to improve the classification of speech data, offering efficient processing times and improved voice quality, particularly through mel-cepstral coefficients.
Once the features are extracted, they are fed into a CNN model. The proposed CNN architecture includes one or more pairs of convolutions and max-pooling layers, which process the input speech signals to classify the corresponding emotions. The model is implemented in MATLAB and evaluated against traditional methods like Linear Prediction Cepstral Coefficient (LPCC) combined with a K-Nearest Neighbor (KNN) classifier.
For performance evaluation, various statistical measures such as accuracy, precision, specificity, recall, sensitivity, error rate, receiver operating characteristics (ROC) curve, area under curve (AUC), and False Positive Rate (FPR) are employed. These metrics help compare the effectiveness of the proposed CNN models against existing methods
A Comprehensive Tool for Legal Document Interpretation and Summarization using Large Language Models
No full textThe proposed system in this paper introduces a user-friendly software solution leveraging cutting-edge AI technology called Large Language Models (LLMs) to simplify the understanding of legal documents and ensure fairness within the legal system. With LLMs at its core, the system offers two primary functions. Firstly, users can upload various legal documents, such as contracts or statutes, and ask questions related to their content. Using sophisticated natural language processing techniques, the system analyses these documents and provides accurate answers, aiding both legal professionals and individuals without legal expertise in navigating complex legal texts effortlessly.
By harnessing the power of LLMs, this software revolutionises how we interact with legal documents. Its advanced capabilities enable users to better understand legal papers and ensure they're fair and transparent. With its user-friendly interface and focus on leveraging LLM technology, the system aims to empower users to make informed decisions and promote fairness and accountability within the legal domai
Next-Gen Financial Facilities: Merging AI and Cloud in Banking
No full textThe monetary administrations industry is experiencing a emotionalmove due to the conversion of manufactured insights and cloudcomputing. This permits banks to supply novel items andadministrations, progress customer encounters, and incrementoperational productivity. This unique examines the ways in whichthese two technologies are combining to convert the keeping moneydivision. AI is making a difference banks make superior choices,streamline strategies, and lower human blunder rates through itspowers in computerization, prescient analytics, and informationinvestigation. Banks may work more rapidly and viably muchobliged to cloud computing, which offers versatile and versatileframework to meet the enormous information handling and capacityneeds of AI-driven frameworks. Personalized encounters and movedforward client benefit are two of AI's fundamental impacts inmanaging an account. Chatbots, virtual collaborators, andsuggestion motors driven by artificial intelligence have made itconceivable for banks to supply customized arrangements to eachspecial client, expanding client bliss and engagement. AIframeworks are able to expect client requests and give pertinentmerchandise and administrations in real-time by assessing giganticdatasets from client intelligent, exchange histories, and outsideadvertise circumstances. Banks have been able to build strongerclient associations and keep a competitive advantage in a showcasethat's changing rapidly since to this customisation. Anothersignificant range of alter is the utilize of AI to chance administrationand extortion discovery. Customary strategies for recognizingfraudulent activity regularly depend on pre-established criteria andhuman assessments, which are difficult and less fruitful inobstructing complex extortion plans. With its machine learningpowers, AI can spot odd patterns, flag conceivable extortion, andmake strides banks' in general security setup
Incorporation of information systems in healthcare centers in the peruvian context: An exploratory systematic review
No full textThe article presents an exploratory systematic review on the incorporation of Information Systems (IS) in healthcare centers in the Peruvian context, highlighting the importance of these technologies in the management of clinical and hospital data. The objective was to analyze the current state of IS implementation in healthcare centers in Peru, identifying the types of integrated systems, the development methodologies adopted, the purpose of the systems, and their impact on medical care. The methodology employed was an exploratory systematic review, in which studies were searched in academic databases such as Scopus. Inclusion criteria were applied, encompassing articles published between 2015 and 2023, in English or Spanish, focused on the implementation of IS in Peruvian healthcare centers. A total of 14 relevant articles were selected. The main findings reveal that the implemented IS include point-of-care medical information systems, electronic health records, and machine learning models. These systems have improved the quality of medical care, optimized hospital processes, and facilitated clinical data management. However, their adoption faces barriers such as a lack of technological infrastructure and resistance to change. In conclusion, the integration of IS in Peruvian healthcare centers has had a positive impact, but challenges remain that require further research and technological adaptation
The Role of Leadership in Project Success: A Quantitative Analysis
No full textThis study explores the relationship between leadership and successful project results by offering a detailed quantitative analysis. The current study focuses on the leadership style that may be attributed to successful delivery and that has a positive relationship with project outcomes. Using a sample of 500 project managers from different industries and the adoption of validated measures to gauge leadership attributes, it follows there are clear success indicators in projects. Overall, we come up with significant strong statistical links of some leader behaviors with the performance variables of a project through correlation and multiple regression analyses. The results indicated strong links of transformational leadership and emotional intelligence with better project outcomes from the strategic decision-making ability of a leader. The research contributes to knowledge by providing empirical data on the critical role that leadership plays in project management and provides relevant practical insights into how the success of projects can be furthered with focused leadership development
Early Prediction of Hyperglycemia Using Cat boost Ensemble Technique
No full textHyperglycemia, characterized by elevated blood glucose levels, is a critical condition that can lead to severe health complications if not detected and managed early. This study explores the application of the Cat Boost ensemble technique for the early prediction of hyperglycemia. Cat Boost, a gradient boosting algorithm that handles categorical features efficiently, is employed to develop a predictive model using a comprehensive dataset comprising patient demographics, medical history, lifestyle factors, and genetic information. The dataset undergoes rigorous preprocessing, including data cleaning, feature engineering, and normalization. The model is trained and validated using an 80-20 train-test split and evaluated through cross-validation to ensure robustness. Key performance metrics such as accuracy, precision, recall, F1-score, and ROC-AUC are utilized to assess the model’s effectiveness. This study demonstrates the potential of the Cat Boost ensemble technique in the early detection of hyperglycemia, offering a valuable tool for healthcare professionals to identify at-risk individuals and implement timely interventions. The proposed model provides 86.15% in prediction of hyperglycemia
The Impact of Post-Quantum Cryptography on Secure Communication Networks: A Review of Current Trends and Future Directions
No full textAs quantum computing advances, traditional cryptographic systems that underpin modern secure communication networks are increasingly at risk. Post-quantum cryptography (PQC) has emerged as a critical solution to secure data against quantum threats. This review examines the impact of PQC on secure communication networks, focusing on current trends, challenges, and future research areas. By exploring recent developments, challenges in implementation, and potential research avenues, we provide a comprehensive overview of PQC's role in safeguarding communication networks in the post-quantum era
Network vulnerability Prediction by using the SVM Technique
No full textThe increasing use of healthcare devices and their communication networks has raised concerns about the security of patient information and the potential for cyber-attacks. This study proposes a machine learning (ML) approach to classify security vulnerabilities in healthcare device communication networks. We collected a dataset of vulnerabilities specific to healthcare devices and applied feature selection and engineering techniques to identify the most relevant features for the classification task. In this paper authors applied the ML techniques and trained several machine learning algorithms, including the Snort algorithm and support vector machines (SVM), to evaluate their effectiveness in this context. We assessed the performance of these models using various evaluation metrics, focusing on accuracy, precision, recall, and F1-score. The results demonstrated that both the SVM and Snort algorithms achieved an accuracy of 94%, a precision of 95%, a recall of 93%, and an F1-score of 94%. These metrics indicate that our ML-based approach is highly effective in identifying security vulnerabilities in healthcare device communication networks. Our approach offers a robust solution for the classification of security vulnerabilities, enabling healthcare providers to identify and prioritize potential threats. This can lead to improved security practices and enhance patient safety by mitigating the risks associated with cyber-attacks on healthcare devices. Our findings suggest that integrating machine learning techniques into security protocols can significantly bolster the defenses of healthcare communication networks
The Role of Digital Narrative Patterns in the Metaverse Era on Human Machine Learning Interaction Systems: A Comparative Analysis of Pre and Post-Interactive Narratives
Get PDFThis study explores the metaverse's intriguing mysteries, including storytelling patterns and interactive narratives' effects on human-computer interactions. This study examines the influence of user-generated tales in the dynamic digital world and evaluates emotional computing models to propose a metaverse-specific framework. The study incorporates concepts from significant works in emotional computing, digital storytelling, and human-computer interaction to improve educational affective computing research. The literature study examines the emotional involvement of digital stories. The article reviews numerous authors' works on emotion-detecting and reacting AI systems. A foundation has been laid for researching metaverse emotions and narrative features. An analytical comparison approach integrates multiple methodologies. Qualitative methods allow for a complete literature review of metaverse user interactions with pre- and post-interactive narratives. Comparative analysis evaluates current emotional computing models to uncover flaws and inform new frameworks. The study's primary focus is comparing story frameworks with emotional computing models to find patterns, similarities, and contrasts. The research shows how metaverse storytelling frameworks have evolved and how user-generated stories affect human-robot relationships. Examining metaverse emotional computing models shows that there are restrictions. Addressing these issues requires a customised approach. Dynamic adaptability, context-aware computing, and a personalised user experience are proposed to improve the metaverse experience. These elements solve the issues and create a more engaging and effective atmosphere. Given the metaverse's growth, this study sheds light on the ever-changing dynamics of digital narratives and emotional computing. The research highlights the vital link between user-generated tales and machine learning systems, which might change digital storytelling. The emotive computing architecture is customised to the metaverse's dynamic and user-centric nature. Amidst the fast expansion of the digital world, it serves as a basis for discipline research and improvement
A Study of Innovative Technologies for Energy-Efficient Enterprise Management of Wireless Heterogeneous Networks in Collaborative Communications
Get PDFCollaborative communication technology has become a popular research area in wireless communications due to its ability to resist varying degrees of channel fading through the collaborative transmission of network nodes. This thesis focuses on energy-efficient collaborative communication systems in increasingly complex environments in heterogeneous wireless networks, with the aim of optimizing energy efficiency and improving user data rates in small areas (e.g., within an enterprise). A brief introduction to the basic technologies of wireless energy-carrying collaborative communication systems is given, summarising relay forwarding strategies, three basic communication models, and energy and information co-transmission reception mechanisms before proposing an ED-OEH relaying protocol at the end of the section that integrates energy classification and opportunity energy harvesting. Immediately afterwards, the heterogeneity of network nodes in terms of computation and storage is pointed out, and a sensor network security protocol based on a hybrid encryption regime is designed. Finally, the problem of intra-enterprise resource allocation and energy efficiency optimization in heterogeneous wireless network scenarios based on deep augmented learning algorithms is investigated. Nature DQN is used as the core algorithm, and the input dimension and loss function in traditional neural networks are improved to reduce the complexity of the algorithm. Experimental results show that the Nature DQN algorithm converges faster than traditional algorithms such as Q-learning, and the energy efficiency ratio can reach up to 300%