检索结果-内蒙古大学图书馆

Comprehensive Integration of Machine Learning, Deep Learning, and Fine-Tuned Large Language Models for Malayalam News Categorization

引用

Procedia computer science 2025年 258卷 3357-3368页

作者： C V Anushka Ardra K John Aarathi Rajagopalan Nair Deepa Gupta G Veena Department of Computer Science and Engineering Amrita School of Computing Bengaluru Amrita Vishwa Vidyapeetham India Department of Computer Science and Applications Amrita School of Computing Amrita Vishwa Vidyapeetham Amritapuri India

The comprehension of news is substantially improved when it is conveyed in an individual’s native language. The increasing volume of news articles, driven by global events, necessitates the effective organization and categorization of vast amounts of information. Dravidian languages, such as Malayalam, characterized by their intricate morphology and diacritical features, pose distinctive challenges in this domain. This research embarks on an innovative initiative to classify Malayalam news articles into four principal categories: business, entertainment, sports, and technology. A synthesis of Deep Learning and Machine Learning methodologies, alongside diverse embedding techniques, is utilized to assess the precision and effectiveness of the categorization endeavor. The ensemble classifiers XGBoost and Stacking, when integrated with the XLM RoBERTa word embedding, attained the highest accuracy rate of 0.91. Similarly, XLM RoBERTa and FastText word embedding techniques, attained a performance with an accuracy rate of 0.93 for almost all the deep learning model. Furthermore, large language models were leveraged, investigating both zero-shot learning and fine-tuning strategies. The fine-tuned BLOOM model reached an accuracy level of 0.89 The research additionally presents a comprehensive examination of the importance of large language models in facilitating the categorization process. Furthermore, the research provides a comprehensive examination of the role of large language models in enhancing the categorization process. By advancing the discipline of Natural Language Processing, this investigation provides significant insights into the utilization of sophisticated technologies for regional news analysis, underscoring their potential to enhance information management and accessibility.

关键词： Natural Language Processing Large Language Models Embeddings Zero-shot learning Malayalam News categorization

来源：评论

学校读者我要写书评

暂无评论

A Way to Design Fog Computing Model for 5G Network using Vanet 4

A Way to Design Fog Computing Model for 5G Network using Van...

引用

4th IEEE International Conference on Advance Computing and Innovative Technologies in engineering, ICACITE 2024

作者： Maruthamuthu, R. Patel, Neeraj Yawanikha, T. Jayasree, S. Alsalami, Zaid Subbarao, S.P.V. Madanapalle Institute of Technology & Science Computer Applications Andhrapradesh Madanapalle India Ies Institute of Technology and Management Ies University Computer Science & Engineering Department M.P Bhopal India Prince Shri Venkateshwara Padmavathy Engineering College India Karpagam Institute of Technology Department of Information Technology Coimbatore641105 India The Islamic University of Babylon College of Technical Engineering Department of Computers Techniques Engineering Babylon Iraq Sreenidhi Institute of Science and Technology Department of Electronics and Communication Engineering Hyderabad India

ISBN: (纸本)9798350360165

Intelligent cities rely heavily on smart automobiles, providing portable computing as well as communication capabilities via a wireless link. By Mobile spontaneous groups, computer fog acts as a middleman between automobiles and the web, giving them access to storing, retrieving and abilities to network. With VANET, which stands processing the request to the vehicle takes fewer minutes since it is a latency-sensitive tech. The most well-known problems with fogged and VANETs are congestion and delays. In this study, we created a priority-based fogged approach to intelligent metro automobile journeys, which lowers fog compute time and postpone, to address such issues. 5G localised Multi-Access edge information workstations was also used to enhance the computer fog infrastructures in meeting timely and The caliber of Level standards, which has greatly reduced the latency and wait. In lieu of the study conducted utilizing simply a web-based design, we achieved a 20% reduction in data time. Additionally, we achieved a 35% reduction in CPU usage as compared to the use of a public cloud platform. © 2024 IEEE.

关键词： 5G mobile communication systems

来源：评论

学校读者我要写书评

暂无评论

Fusion of Frequency Domain Features of Face and Iris Traits for Person Identification

引用

Journal of The Institution of Engineers (India): Series B 2021年第5期102卷 987-996页

作者： Kagawade, Vishwanath C. Angadi, Shanmukhappa A. Department of Computer Applications Basaveshwar Engineering College Bagalkot India Department of Computer Science and Engineering Centre for Post Graduate Studies VTU Belagavi India

In this study, we present a new approach for person recognition through feature level fusion of face and iris traits. The strength of the multimodal biometric system relies much more on the consistency to determine relevant features from the individual biometric traits and fusion technique employed for integration. Hence, this paper presents a new feature extraction approach for a face–iris multimodal biometric system. The face and iris feature extraction is performed by employing Polar Fast Fourier Transform to obtain textural information in diverse scales and orientations of the face and eye images. The Canonical Correlation Analysis method is employed to integrate face and iris traits features, which reduces the new feature vector length than the input feature vectors. The proposed approach is capable to obtain and improve the discriminative power for person recognition from multi-feature data of face and iris traits. The effectiveness of the proposed model is illustrated through the use of both artificial and genuine face and iris multimodal databases, along with comparisons involving existing state-of-the-art techniques. © 2021, The Institution of Engineers (India).

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Facial Expression Recognition Using Multi-Block Deep CNN 5

Facial Expression Recognition Using Multi-Block Deep CNN

引用

5th International Conference on Circuits, Control, Communication and Computing, I4C 2024

作者： Priyadarshini, V. Srinivasulu Reddy, U. Reddy Chirra, Venkata Rami Mrudula, M. Suneetha, M. National Institute of Technology Dept. of Computer Science & Engineering Andhra Pradesh India Machine Learning & Data Analytics Lab National Institute of Technology Dept. of Computer Applications Tiruchirappalli India School of Computer Science & Engineering VIT-AP University Amaravathi India Malla Reddy College of Engineering Department of CSE Telangana India Independent Researcher Amaravati India

ISBN: (纸本)9798331528539

Emotions play a crucial role in decision-making, moral judgments, and other cognitive processes. The goal of this work is to identify people's facial emotions from face images. Facial expression recognition (FER), a highly diverse field with many commonalities between classes, is a challenging problem. The capabilities of current FER systems are limited in this regard. In this work, a multi-block Deep Convolutional Neural Networks (DCNN) model was proposed to identify facial expressions. Five blocks with different computational components were defined in a multi-block DCNN to extract the unique characteristics from input images. The second model (DCNN-SVM) is developed with an ensemble of SVM classifier to improve predictability and stability. The third model (DCNN-MV) is based on the majority voting with 4 classifiers SVM, KNN, Random Forest, and Decision Tree. The CK+ dataset was utilized and expanded via image data augmentation to enhance model performance and generalization. By adjusting hyperparameters, the suggested DCNN model's accuracy was investigated. The DCNN model reports an accuracy score of 97.71%. Both DCNN-SVM and DCNN-MV models achieved 97.79% accuracy. On the CK+ dataset, the proposed models demonstrated good accuracy in identifying facial expressions. When compared to existing models, the proposed model outperformed existing models in terms of accuracy. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Surveillance and Intruder Detection using Pyramid Vision Transformer You Only Look Once Enhanced with Gradient Weighted Class Activation Mapping 4

Surveillance and Intruder Detection using Pyramid Vision Tra...

引用

4th IEEE International Conference on Mobile Networks and Wireless Communications, ICMNWC 2024

作者： Arunachalam, Sripriya Almusawi, Muntather Rafikiran, Shaik. Sivagami, S. Praveena, Hirald Dwaraka Srm Institute of Science and Technology Faculty of Science and Humanities Department of Computer Science and Applications Vadapalani City Campus Chennai India College of Technical Engineering The Islamic University of Al Diwaniyah Department of Computers Techniques Engineering Al Diwaniyah Iraq Sri Venkateswara College of Engineering Department of Electrical & Electronics Engineering Tirupati India New Prince Shri Bhavani College of Engineering and Technology Department of Chemistry Chennai India Department of Electronics and Communication Engineering Tirupati India

ISBN: (纸本)9798350352931

In this era, wireless networks are economically feasible and flexible to establish in healthcare, defense, surveillance, and traffic monitoring, fire detection. Effective intruder detection systems utilize advanced technologies like Artificial Intelligence (AI), sensors, and cameras. Traditional approaches for surveillance and intruder detection had faced several challenges which include false alarms, high costs, privacy concerns, and vulnerabilities. To overcome these challenges, a Gradient-weighted Class Activation Mapping (Grad CAM) with Pyramid Vision Transformer-You Only Look Once Enhanced (PViT-YOLOE) namely GRADCAM-PViT-YOLOE is proposed to detect intruders in public places using images and videos. For preprocessing, noise reduction, image resizing, and image segmentation are considered to reduce false positives, enhance memory efficiency, and to reduce complexity. Then, PViT-YOLOE is considered for feature extraction to enhance detection of small objects, enhance robustness to variations. At last, Grad CAM is employed as classifier to detect intruders in public places with improved interpretability, robustness to noise, and flexibility. The results demonstrate the efficacy of the proposed GRADCAM-PViT-YOLOE in terms of accuracy (99.25%), precision (96.75%) recall (89.82%) and F1-Score (87.92%) when compared to the existing method like Faster Regions Convolutional Neural Networks (FR-CNN). © 2024 IEEE.

关键词： Noise abatement

来源：评论

学校读者我要写书评

暂无评论

Optimizing Inventory Management using Demand Forecasting based Ensemble Stacking Model with Hyperparameter Tuning 1

Optimizing Inventory Management using Demand Forecasting bas...

引用

1st International Conference on Intelligent Systems and Computational Networks, ICISCN 2025

作者： Pavithra, V. Abbas, Haayder M. Manjula, B.M. Chidambaram, P.K. Jothilakshmi, S. Srm Institute of Science and Technology Faculty of Science and Humanities Department of Computer Science and Applications Ramapuram Chennai India The Islamic University of Al Diwaniyah College of Technical Engineering Department of Computers Techniques Engineering Al Diwaniyah Iraq Nitte Meenakshi Institute of Technology Department of Electronics and Communication Engineering Bengaluru India New Prince Shri Bhavani College of Engineering and Technology Department of Mechanical Engineering Chennai India Dhanalakshmi Srinivasan College of Engineering and Technology Department of Information Technology Mamallapuram India

ISBN: (纸本)9798331529246

In recent days, the effective inventory management relies on accurate demand forecasting for effective management of business balance stocks. However, the traditional forecasting methods often struggled with different complexities such as variability, overstocking and wasted resources. In this research, the Ensemble Stacking Model with Hyperparameter Tuning (En-Stacking-HT) is proposed for effective optimized inventory management. Initially, the data is collected from M5 accuracy completion dataset and further preprocessed with data imputation, z-score normalization. These preprocessing techniques handles the missing values in data and standardizes the data in significant range. Then, Seasonal and Trend Decomposition (STL) is employed for extracting seasonality and trend features from preprocessed data by using Locally Estimated Scatterplot Smoothing (LOESS). Finally, the proposed En-Stacking-HT is utilized for inventory management prediction and optimized through hyperparameter tuning for predicting effective optimized inventory management for demand forecasting. From the results, the proposed En-Stacking-HT attained better results in terms of Mean Absolute Error (MAE) (16.82), Root Mean Square Error (RMSE) (22.85) and Mean Absolute Percentage Error (MAPE) (4.97%) when compared to existing Random Forest (RF). © 2025 IEEE.

关键词： demand forecasting ensemble stacking model hyperparameter tuning optimized inventory management seasonality and trend decomposition

来源：评论

学校读者我要写书评

暂无评论

Statistical Models for Bacterial Propagation in Human Respiratory Tract

Statistical Models for Bacterial Propagation in Human Respir...

引用

2024 International Conference on Communication, computer sciences and engineering, IC3SE 2024

作者： Jaiswal, Raghevendra Maurya, Rahul Singh, Rajneesh Agarwal, Ruchi Singh, Shyam Pratap Pandey, Vijay Kumar Galgotias College Of Engineering & Technology Department Of Electronics & Communication Engineering Greater Noida India Sharda University India Department Of Computer Science & Applications Sharda School Of Engineering & Technology Greater Noida India Nagpur Campus Pune India Noida Institute Of Engineering & Technology Department Of Electronics & Communication Engineering Greater Noida India

ISBN: (纸本)9798350366846

Diffusive Molecular Communication (DMC) enabled bacterial propagation within biological systems is employed in different medical applications. In the other words, DMC is being used to study the propagation of bacteria inside the human body. However, DMC and thus the bacterial propagation model depend on the type of deployment used. Various deployment geometries including Rectangular, Spherical and Cylindrical. In this research, a DMC system within a cylindrical biological deployment is developed for the human respiratory tract. It is expected that the information molecules diffusing across the fluid medium would degrade and be influenced by flow. Further, the Cylindrical Biological Deployments (CBDs) offer versatility in representing different tissues or organs. Therefore, in this research paper different statistical models are derived and analyzed under CBD deployment using bacterial concentration f(y) In particular, statistical models for bacterial propagation in terms of probability (Pb), non-binding probability (Pnb), evade probability (Pe(y)), and the probability of binding rate (Pb(y)) are presented. Also, presented models are analyzed numerically using simulation. The numerical simulation demonstrates excellent agreement with the theoretical foundation. The result has been simulated in MATLAB-20. © 2024 IEEE.

关键词： Fading channels

来源：评论

学校读者我要写书评

暂无评论

Comparing Collaborative Filtering Algorithms in AI-Powered Recommendation Systems for E-Commerce 4

Comparing Collaborative Filtering Algorithms in AI-Powered R...

引用

4th International Conference on Sentiment Analysis and Deep Learning, ICSADL 2025

作者： Aamer, Najmuddin Mittal, Tanusha Muthusamy, Nantha Kumar Battula, R. S. S. Raju Ajith, P.S. Soudagar, ManzooreElahi M. Bharat Ratna Indira Gandhi College of Engineering Department of Computer Science and Engineering Maharashtra Solapur410003 India Graphic Era Deemed to Be University Department of Computer Science and Engineering Maharashtra Dehradun410003 India OntrailInc Dba Uprise San FranciscoCA94104 United States Aditya University Department of Computer Applications Andhra Pradesh Surampalem533432 India Brindavan College of Engineering Department of Mathematics Karnataka Bengaluru560063 India Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Department of Vlsi Microelectronics Tamil Nadu Chennai602105 India

ISBN: (纸本)9798331523923

The quick growth of e-commerce has brought attention to how important efficient recommendation systems are to enhancing user experience and accomplishing business objectives. This paper investigates the effectiveness of collaborative filtering algorithms in AI-powered recommendation systems, with a focus on user-based collaborative filtering (UBCF). A comprehensive method is employed to guarantee consistency across different rating scales, beginning with data normalisation to standardise user-item interaction matrices. To decrease dimensionality and maintain the most significant patterns of user-item interaction, Principal Component Analysis (PCA) is employed for feature selection. The classification system uses similarity measures and UBCF to predict user preferences based on historical data. The algorithms are evaluated on key metrics like precision, recall, and RMSE using publically available datasets. The outcomes demonstrate the impact of preprocessing and dimensionality reduction on UBCF's performance as well as its efficacy in providing customised recommendations. This research contributes to the improvement of collaborative filtering algorithms for e-commerce systems, which will raise customer satisfaction and enable scalability in dynamic online markets. © 2025 IEEE.

关键词： Collaborative filtering

来源：评论

学校读者我要写书评

暂无评论

Machine Learning Based ADMET Prediction in Drug Discovery 4

Machine Learning Based ADMET Prediction in Drug Discovery

引用

4th IEEE Global Conference for Advancement in Technology, GCAT 2023

作者： Bhavitha Lekshmi Prasad, P.S. Ani, R. Deepa, O.S. Department of Computer Science and Applications Amrita School of Computing Amrita Vishwa Vidyapeetham Amritapuri India Department of Mathematics Amrita School of Engineering Amrita Vishwa Vidyapeetham coimbatore India

ISBN: (纸本)9798350305258

A drug is a substance or compound with the ability to treat or prevent disease in people or animals. Drug discovery is the process of identifying prospective medications for certain ailments, and it can take several years to accomplish. However, the introduction of machine learning has simplified the procedure. Virtual screening's major goal is to minimize the virtual compound space of biological molecules. To do this, we developed the VIRTUAL SCREENING TOOL, a virtual molecular descriptor calculator that can analyze molecules and assess their drug-likeness based on specific rules. We used Django with PyCharm, a Python IDE, to create a molecular description calculator, fingerprint generator, and prediction interface. The tool begins by inputting SMILES, which is the chemical structure of the molecule, and outputs molecular descriptors, which are the properties of the molecules, including their physicochemical properties, absorption, distribution, metabolism, lipophilicity, excretion, toxicity, drug-likeness rules, and fingerprint generation. The outputs are available in CSV format for download. © 2023 IEEE.

关键词： Drug Fingerprint Physiochemical property Prediction SMILES

来源：评论

学校读者我要写书评

暂无评论

Blockchain for 5G-Enabled IoHT—A Framework for Secure Healthcare Automation

Blockchain for 5G-Enabled IoHT—A Framework for Secure Healt...

引用

International Conference on Data Analytics and Management, ICDAM 2022

作者： Alam, Md Imran Ahmad, Md Oqail Siddiqui, Shams Tabrez Khan, Mohammad Rafeek Khan, Haneef Qidwai, Khalid Ali Department of Computer and Network Engineering Jazan University Jazan45142 Saudi Arabia Department of Computer Applications B.S Abdur Rahman Crescent Institute of Science and Technology Tamil Nadu Chennai600048 India Department of Computer Science Jazan University Jazan45142 Saudi Arabia

ISBN: (纸本)9789811976148

Medical care has evolved into a lifeline for many people;due to this, there is an explosion in the field of medical big data. Wearable technology based on the Internet of Things (IoT) is being used by healthcare professionals to speed up diagnosis and treatment. There are now numerous sensors and devices, which are connected to the Internet. Internet of HealthThings (IoHT) can be used in many different ways in healthcare. Sensors and remote-monitored medical devices can help patients get more accurate and efficient diagnostics. This paper focuses on healthcare automation using the Internet of Things (IoT)-enabled devices. IoT-driven healthcare models can keep patients healthy and safe, enabling doctors to provide better care. In addition, healthcare frameworks significantly reduce healthcare costs and improve performance. Blockchain systems are transparent and thus safe to store and share private information. Data collected by medical devices attached to the body is kept in a cloud server alongside relevant medical reports as part of the proposed framework. The use of blockchain and 5G technology means that patients’ data can be securely transmitted at high speeds and with rapid response times. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Internet of things

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：