检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Khan, Muhammad Danyal Ali, Raheem Aziz, Arshad Department of Cyber Security Pakistan Navy Engineering College National University of Science and Technology Karachi Pakistan Department of Computer Science Bahria University Karachi Campus Karachi Pakistan

Call Centers are an effective medium for businesses, organizations and even government institutions to interact with their clients despite availability of mobile apps and websites. These centers have huge amount of audio data which can be used for achieving valuable business insights. This requires transcription of phone calls which is manually tedious task to perform. An effective Automated Speech Recognition system can accurately transcribe these calls, making it easier to search through call history for specific context and content with very less time and manual effort. With text transcription available calls can be automatically monitored, improving QoS through keyword search and sentiment analysis. ASR for Call Center requires more robustness as telephonic environment are generally noisy. Moreover, there are many low-resourced languages that are on verge of extinction which can be preserved with help of Automatic Speech Recognition technology. Urdu is the 10th most widely spoken language in the world, with 231,295,440 worldwide still remains a resource constrained language in ASR. Regional call-center conversations operate in local language, with a mix of English numbers and technical terms generally causing a "code-switching" problem. Hence, this paper describes an implementation framework of a resource efficient Automatic Speech Recognition/ Speech to Text System in a noisy call-center environment using Chain Hybrid HMM and CNN-TDNN for Code-Switched Urdu Language. Using Hybrid HMM-DNN approach allowed us to utilize the advantages of Neural Network with less labelled data. Adding CNN with TDNN has shown to work better in noisy environment due to CNN’s additional frequency dimension which captures extra information from noisy speech, thus improving accuracy. We collected data from various open sources and labelled some of the unlabelled data after analysing its general context and content from Urdu language as well as from commonly used words from other langua

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Neuromorphic-Driven Agentic AI for Autonomous Decision-Making Systems

Neuromorphic-Driven Agentic AI for Autonomous Decision-Makin...

引用

Mobile Networks and Wireless Communications (ICMNWC), IEEE International Conference on

作者： Manjunath Kamath K Samata Mehta.S Akshaya H. L Shilpashree N Girish Jadhav Abhijit Mitra Department of Cyber Security NMAM Institute of Technology NITTE (Deemed to be University) Karkala India Department Of CSE (IOT & Cyber Security including Block Chain Technology) Brindavan College of Engineering Bengaluru India Department of Computer Science Sarada Vilas College Mysuru India Department Of Electrical Engineering Faculty of Engineering and Technology Parul institute of Engineering and Technology Parul University Post Limda India Symbiosis Centre for Advanced Legal Studies and Research (SCALSAR) Symbiosis Law School Symbiosis International (Deemed University) Pune India

ISBN: (数字)9798331528348

ISBN: (纸本)9798331528355

Agentic AI represents a paradigm shift in the development of intelligent systems capable of adaptive and proactive interactions in dynamic and complex environments. By integrating reinforcement learning (RL) with cognitive frameworks, Agentic AI goes beyond traditional rule-based and reactive models, enabling autonomous systems to make informed decisions, anticipate future states, and learn from experience. This paper explores the theoretical foundations and practical applications of Agentic AI, highlighting its potential to transform a variety of fields, including robotics, autonomous driving, finance, and healthcare. Through a detailed review of state-of-the-art research, we illustrate how cognitive architectures such as ACT-R and Soar, combined with advanced RL techniques like Deep Q-Networks (DQN) and Proximal Policy Optimization (PPO), contribute to the development of AI agents with human-like reasoning and decision-making capabilities. Experimental results demonstrate that Agentic AI significantly outperforms conventional AI approaches in terms of adaptability, learning efficiency, and decision accuracy. The findings suggest that Agentic AI offers a robust framework for creating intelligent systems capable of complex problem-solving, long-term planning, and proactive behavior, paving the way for the next generation of AI-driven applications.

关键词： Adaptation models Adaptive systems Decision making Finance Medical services Reinforcement learning Cognition Autonomous vehicles Robots Optimization

来源：评论

学校读者我要写书评

暂无评论

Non-Linearly Weighted Pheromone Updating for Ant Colony Optimization

Non-Linearly Weighted Pheromone Updating for Ant Colony Opti...

引用

2024 IEEE International Conference on Systems, Man, and cybernetics, SMC 2024

作者： Qiu, Ying-Han Yang, Qiang Li, Jian-Yu Jia, Ya-Hui Wang, Zi-Jia Gao, Xu-Dong Lu, Zhen-Yu Zhang, Jun College of Artificial Intelligence Nankai University Tianjin China School of Computer Science and Cyber Engineering Guangzhou University Guangzhou China School of Artificial Intelligence Nanjing University of Information Science and Technology Nanjing China School of Future Technology South China University of Technology Guangzhou China Hanyang University Department of Electrical and Electronic Engineering Ansan Korea Republic of

ISBN: (纸本)9781665410205

Ant Colony Optimization (ACO) has witnessed great success in tackling the Traveling Salesman Problem (TSP). In ACO, ants involved in the pheromone update play pivotal roles in its optimization effectiveness. Along this road, this paper designs an ant selection mechanism along with a non-linear weight method for ACO to update the pheromone effectively, leading to a novel ACO, called NLW-ACO. Particularly, NLW-ACO leverages the fitness values of ants to assign each ant a selection probability. Then, it adaptively chooses ants for pheromone update. Subsequently, a nonlinear weight is assigned to each selected ant based on its fitness value to update the pheromone matrix. Resultantly, better ants have higher selection probabilities and larger weights to take part in the pheromone update. This leads to that NLW-ACO compromises search convergence and search diversity appropriately to seek for the optimum. Experiments have been carried out on 10 TSP instances of diverse scales. The experimental findings substantiate that NLW-ACO significantly outperforms the 5 typical ACO methods, especially on large-scale TSP problems. © 2024 IEEE.

关键词： Ant colony optimization

来源：评论

学校读者我要写书评

暂无评论

Compressing Time Series Towards Lightweight Integrity Commitments 8

Compressing Time Series Towards Lightweight Integrity Commit...

引用

8th South-East Europe Design Automation, computer engineering, computer Networks and Social Media Conference, SEEDA-CECNSM 2023

作者： Katsika, Angeliki Papageorgiou, Konstantinos Fakis, Alexandros Kakarountas, Athanasios Andritsopoulos, Fotis Plagianakos, Vassilis Spathoulas, Georgios University of the Aegean Department of Information and Communication Systems Engineering Samos Greece ITrack Athens Greece University of Thessaly Department of Computer Science and Biomedical Informatics Lamia Greece Department of Information Security and Communication Technology Gjovik Norway

ISBN: (纸本)9798350383676

In the current era of big data, the generation and preservation of time series data has seen widespread adoption in various sectors such as trade, healthcare, and industry. At the same time, data have become a prized asset, as we are witnessing an unprecedented explosion in the volume of data generated and collected. This includes data from various sources, such as sensors, social media, and transactions, offering a competitive edge to organizations that harness the power of data, leading them to informed decisions and rapid responses to changing conditions. However, it is essential to responsibly manage data, addressing privacy and security concerns to ensure their integrity and availability, while efficiently managing the cost of storage and operations. To address these challenges, there is a compelling de-mand for methods that compress these time series while ensuring that the compressed versions faithfully represent the original data and remain secure against tampering or unauthorized alterations. This paper introduces a novel approach to condensing time-series data into lightweight integrity commitments, leveraging polynomial vector commitments as a key component. These integrity commitments are integrated with blockchain technology, serving as a cornerstone in guaranteeing data integrity and authenticity, particularly in supply chain scenarios where trust and transparency are of the utmost importance. © 2023 IEEE.

关键词： Time series

来源：评论

学校读者我要写书评

暂无评论

RVAFM: Re-parameterizing Vertical Attention Fusion Module for Handwritten Paragraph Text Recognition

arXiv

引用

arXiv 2025年

作者： Zheng, Jinhui Liu, Zhiquan Si, Yain-Whar Li, Jianqing Zhang, Xinyuan Li, Xiaofan Huang, Haozhi Gong, Xueyuan School of Intelligent Systems Science and Engineering Jinan University Guangdong Province China College of Cyber Security Jinan University Guangdong Province China Department of Computer and Information Science Faculty of Science and Technology University of Macau China School of Computer Science and Engineering Macau University of Science and Technology China

Handwritten Paragraph Text Recognition (HPTR) is a challenging task in computer Vision, requiring the transformation of a paragraph text image, rich in handwritten text, into text encoding sequences. One of the most advanced models for this task is Vertical Attention Network (VAN), which utilizes a Vertical Attention Module (VAM) to implicitly segment paragraph text images into text lines, thereby reducing the difficulty of the recognition task. However, from a network structure perspective, VAM is a single-branch module, which is less effective in learning compared to multi-branch modules. In this paper, we propose a new module, named Re-parameterizing Vertical Attention Fusion Module (RVAFM), which incorporates structural re-parameterization techniques. RVAFM decouples the structure of the module during training and inference stages. During training, it uses a multi-branch structure for more effective learning, and during inference, it uses a single-branch structure for faster processing. The features learned by the multi-branch structure are fused into the single-branch structure through a special fusion method named Re-parameterization Fusion (RF) without any loss of information. As a result, we achieve a Character Error Rate (CER) of 4.44% and a Word Error Rate (WER) of 14.37% on the IAM paragraph-level test set. Additionally, the inference speed is slightly faster than VAN. © 2025, CC BY.

关键词： Parameterization

来源：评论

学校读者我要写书评

暂无评论

Optimizing Phishing URL Detection: A Comparative Analysis of Feature Counts and Training Times 5

Optimizing Phishing URL Detection: A Comparative Analysis of...

引用

5th International Conference on Intelligent Computing, Communication, Networking and Services, ICCNS 2024

作者： Alajlouni, Moayyad Tashtoush, Yahya Dossari, Saud Alhajaj Al. Darwish, Omar Jordan University of Science and Technology Department of Computer Science Irbid Jordan Prince Sattam bin Abdulaziz University Department of Electrical Engineering Wadi Addwasir Saudi Arabia Eastern Michigan University Information Security and Applied Computing Department YpsilantiMI48197 United States

ISBN: (纸本)9798350354690

Phishing attacks are the biggest cybersecurity threats in the digital world. Attackers exploit users by impersonating real, authentic web sites to obtain sensitive information such as passwords and bank statements. One common technique used in these attacks is malicious URLs. These malicious URLs mimic legitimate URLs, misleading users into interacting with malicious websites. This practice, URL phishing, presents a big threat to internet security, emphasizing the need for advanced detection methods. This paper presents a method for detecting malicious URLs using statistical features extracted from n-grams. These n-grams are extracted from the hexadecimal representation of URLs. We used Explainable AI (XAI) to explore the extracted features and evaluate their importance and role in phishing detection. A key advantage of our method is its ability to reduce the required features and the training time by using fewer features after applying XAI techniques. Our technique only uses statistical features extracted from n-grams and the n-gram itself, without requiring any high-level features. We applied XAI techniques, SHapley Additive exPlanations (SHAP), and Local Interpretable Model-agnostic Explanations (LIME). Through the explanation provided by XAI methods, we were able to determine the most important features in our feature set, enabling a reduction in feature count. Using fewer features (4, 7, 10, 13, 15), we got good accuracy compared to the 41 features and 87 features used in the original experiment and reduced the models' training times and complexity. This research aimed to enhance phishing URL detection by optimizing feature selection and analyzing the impact on training times. Our findings show the importance of using minimal statistical features to identify malicious URLs effectively and efficiently. © 2024 IEEE.

关键词： Phishing

来源：评论

学校读者我要写书评

暂无评论

Price Prediction of Digital Currencies using Machine Learning 10

Price Prediction of Digital Currencies using Machine Learnin...

引用

10th IEEE International Conference on Data Science and Advanced Analytics, DSAA 2023

作者： Dwivedi, Ashutosh Dhar Dutta, Subhayu Adhikary, Subhrangshu Pedersen, Jens Myrup Aalborg University Cyber Security Group Department of Electronic Systems Copenhagen Denmark Dr. B.C. Roy Engineering College Department of Computer Science & Engineering West Bengal Durgapur713206 India Spiraldevs Automation Industries Pvt. Ltd. Department of Research & Development West Bengal Raiganj733123 India

ISBN: (纸本)9798350345032

Cryptocurrencies have gained immense significance and popularity in recent times. With thousands of digital currencies available, selecting the right one can be challenging for users. In the financial sector, accurately predicting future prices is crucial for profitable investments in digital currencies. However, price prediction in this realm poses unique challenges, as it lacks physical goods or services as the basis, unlike stock prices. Machine learning emerges as a pivotal tool for addressing this challenge and plays a vital role in price prediction. This research analyzes five prominent currencies - Monero, Bitcoin, Ethereum, IOTA, and Zcash - employing five models: SVR, LRG, Huber, RANSAC, MLP, and AdaBoost. The experimental results demonstrate promising outcomes, showcasing the ability to predict digital currency prices with an impressive R2 score of 1.0 for specific machine learning algorithms. This advancement opens new avenues for informed decision-making and profitable ventures in the dynamic world of digital currencies. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Classification and Labeling Techniques of Educational Resources Based on BERT-CNN model

Classification and Labeling Techniques of Educational Resour...

引用

Emerging Smart Computing and Informatics (ESCI), Conference on

作者： Viswanathan.M.P Noor Kaylan Hamid V Rajyalakshmi Neelam Sharma Sharmeen Izzat Hassan S Muthuperumal Hindusthan College of Engineering and Technology Coimbatore India Department of Computer Education Al Kitab University Kirkuk Iraq Department of CSE (Cyber Security) Vignana Bharathi Institute of Technology India School of Physical Education Lovely Professional University Punjab India Department of Computer Engineering College of Engineering Knowledge University Erbil Iraq Department of Mathematics K.Ramakrishnan College of Engineering Trichy India

ISBN: (数字)9798331515683

ISBN: (纸本)9798331515690

The key factor in promoting accessibility and improving the educational experience is accurate classification and labelling of educational materials. Research paper presents a novel approach to educational material classification that makes use of BERT-CNN, a hybrid model that combines and CNNs. The limitations of conventional approaches are addressed by research methodology, which seamlessly integrates a contextual understanding of BERT with CNN's extraction capabilities to address the prior difficulty. It has frequently been difficult for prior techniques to properly capture category elements and comprehend contextual complexities in educational text classification. Here, research leverage the pre-trained contextual embeddings of BERT to enable research model to capture complicated semantics in educational texts, and the CNN architecture successfully extracts categorical features from these embeddings. By combining these two potent models, research is able to capture the logical and structural characteristics of educational resources and create a thorough knowledge of them. Furthermore, research implementation in Python makes use of well-known deep learning packages like TensorFlow and PyTorch, guaranteeing repeatability and accessibility for practitioners and researchers. Research suggested model performs very well, achieving 99% classification accuracy through extensive manipulation and analysis of datasets on various educational resources. The findings demonstrate the efficiency and resilience of research approach in correctly categorizing and labelling instructional materials, underscoring its significance in the management of educational content and strategic enhancement.

关键词： Accuracy Scalability Text categorization Semantics Feature extraction Robustness Labeling Ensemble learning Random forests Context modeling

来源：评论

学校读者我要写书评

暂无评论

Data Analytics for Identifying and Reducing Hospital Readmissions 3

Data Analytics for Identifying and Reducing Hospital Readmis...

引用

3rd International Conference on Advancement in Computation and computer Technologies, InCACCT 2025

作者： Vidhya, K. Srinu, N. Ashok, D. Rajasekaran, Arun Sekar Kumar, Ashok S. Rajasekaran, U. Division of Data Science and Cyber Security School of Computer Science and Technology Karunya Institute of Technology and Sciences Coimbatore India Department of Computer Science Engineering QIS College of Engineering and Technology Ongole India Department of AIMLDS QIS College of Engineering and Technology Ongole India Department of Electronics and Communication Engineering SR University Telangana Warangal India Department of Electronics and Communication Engineering Karpagam Academy of Higher Education Coimbatore India Department of Electronics and Communication Engineering Builders Engineering College Nathakadaiyur India

ISBN: (纸本)9798331543891

Data analytics examines healthcare data to identify patterns and personalize care plans ultimately by reducing hospital readmissions. Readmissions after discharge increase healthcare costs, lower quality, and dissatisfied patients. To overcome these issues, a novel Data analytics approach has been proposed to reduce hospital readmissions. This inquiry begins with a complete dataset of clinical data, patient demographics, and Electronic Health Records (EHRs). Initially, the data gathered from the clinical data, patient demographics, and electronic health records are fed to the preprocessing module for transforming raw healthcare data into a structured form for analysis. After preprocessing, an Exploratory Data Analysis (EDA) is performed which simplifies the initial insights and hospital readmission risk factor identification. The most essential features are selected from the analyzed outcomes of EDA for readmission prediction. After feature selection, the development of hospital readmission models uses three data analytics methods. At first, the Machine Learning (ML) based methods predict the readmission status and the sequential relationships and anticipate readmission risk over time are captured using a neural network. Finally, the patient groups with comparable readmission trends are identified through hierarchical clustering. After data analytics, those outcomes are fed to the real-time patient monitoring system to reveal readmission factors such as discharge planning and timely drug management. The proposed method is validated by using healthcare datasets. In comparison, the proposed method achieves an accuracy of 97% whereas, the predictive modeling, risk factor, care process, and alert method achieves 85%, 78%, 90%, and 83% respectively. © 2025 IEEE.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Research on Distributed Sensor Network Faulty Recovery Method based on Attention Mechanism 3

Research on Distributed Sensor Network Faulty Recovery Metho...

引用

3rd International Conference on Artificial Intelligence and computer engineering, ICAICE 2022

作者： Xing, Zheng Liu, Boya Guan, Haochen Zhang, Yiran Zhang, Ying Chengdu University of Technology The College of Nuclear Technology and Automation Engineering Sichuan Chengdu China Chengdu University of Technology College of Environment and Civil Engineering Sichuan Chengdu China Zhengzhou University of Light Industry International Education College Henan Zhengzhou China Hebei University School of Cyber Security and Computer Hebei Baoding China Northeast Petroleum University School of Electrical & Information Engineering Heilongjiang Daqing China

ISBN: (纸本)9781510663473

Faulty recognition and recovery is extremely essential component for existing sensor networks. However, current investigations concentrates on the recognition process and develop abundant advanced identification methods and existing recovery method is almost based on the multiple layers of neural network to achieve the recovery of sensor network, which would result in vast computation costs. Attention mechanism is proposed for neural network to focus some important neural nodes and perform high accuracy for training process with insignificant computation cost. In this article, we propose a novel neural network to recover the fault sensor network with attention mechanism. Thus, the developed mechanism can achieve rapid recovery of fault sensors network with reasonable computation costs and acceptable performance. From extensive experimental result, we can conclude that our proposed method can strictly process the fault sensor network and compared with existing fault sensor network recovery methods to illustrate the effective performance on recovery accuracy with reasonable costs of our proposed method. © 2023 SPIE.

关键词： Sensor networks

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：