检索结果-内蒙古大学图书馆

2025 International Conference on Computing for Sustainability and Intelligent Future, COMP-SIF 2025

作者： Kumar, B. P. Pradeep Ravi Kumar, J. Biswas, Arpita Hafeeza, Bi Bi Dhaanya, D. Visvesvaraya Technological University Department of Computer Science Design Atria Institute of Technology Karnataka Bangalore India Visvesvaraya Technological University Dept. of Computer Science Engineeering Atria Institute of Technology Karnataka Bangalore India

ISBN: (纸本)9798331538538

Emotion categorization is used in domains like customer service, social media content moderation, and interaction between humans and computers. Classification of emotion from text is a fundamental issue in NLP, or natural language processing. This study examines both advanced deep learning models (LSTM networks with embedding layers) and traditional machine learning techniques (Naive Bayes, Logistic Regression, SVM, Random Forest) for the classification of the six primary emotions: joy, sorrow, anger, fear, love, and surprise. The study uses a preprocessed emotion-tagged text dataset, and metrics including accuracy, precision, recall, and F1-score are utilized to evaluate the models. LSTMs perform better than conventional models, which find it difficult to grasp language subtleties and contextual correlations, by processing sequential data and contextual information with ease. © 2025 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Diagnosis of Breast Cancer Using Vision Transformers (ViT)

Diagnosis of Breast Cancer Using Vision Transformers (ViT)

引用

2025 International Conference on Computing for Sustainability and Intelligent Future, COMP-SIF 2025

作者： Ravikumar, J. Prithvi Prabhu Pani, V. Tejushree, R. Varshitha, Sunnampalli Atria Institute of Technology Department of Computer Science and Engineering Bangalore India

ISBN: (纸本)9798331538538

Breast cancer is one of the leading causes of death in women. Regular testing is of paramount importance since early detection ensures more treatment options and better prognosis rate. Biopsies, x-rays and ultrasounds are some of the diagnostic tools used to detect breast cancer. Out of these, ultrasounds being portable and ionizing radiation free provide the most feasible and harmless solution in areas which cannot accommodate expensive equipment that require rigorous safety standards to be met. Our detection system utilizes Vision Transformer (ViT) technology, i.e., transfer learning to classify tumors of the breast into malignant, benign and normal categories. Data enhancement has been done using Contrast Limited Adaptive Histogram Equalization (CLAHE). Furthermore, we have categorized them based on BI-RADS criteria for greater accuracy and precision scores. We have chosen ViT over the more commonly used Convolution Neural Networks (CNN) since they offer a more flexible approach, capturing global relationships in images, and you can use it for larger datasets. © 2025 IEEE.

关键词： Ultrasonic imaging

来源：评论

学校读者我要写书评

暂无评论

Enhanced perceptual wavelet packet features for spontaneous Kannada sentence recognition under uncontrolled conditions

引用

International Journal of Speech technology 2025年第1期28卷 153-174页

作者： Shanthamallappa, Mahadevaswamy Pradeep Kumar, B.P. Department of Electronics and Communication Engineering Vidyavardhaka College of Engineering Mysuru Visvesvaraya Technological University Karnataka India Department of Computer Science and Design Atria Institute Of Technology Bengaluru Visvesvaraya Technological University Karnataka India

A Robust Automatic Speech Recognition (ASR) system is proposed through a hybrid combination of Perceptual Wavelet Packet features, Deep Neural Network-Hidden Markov Model (DNN-HMM) acoustic models, and n-gram language models for recognizing Spontaneously spoken Kannada sentences. Two Experiments are performed on own Kannada speech corpus and the widely recognized TIMIT corpus. First one is the experiment on the clean speech datasets and the second one is on the speech degraded by noise data chosen from Noise-US dataset. The proposed system performance is evaluated through the metric Word Error Rate (WER). A comparative study on the performance analysis is presented. The results of the experimental investigations achieved an average reduction in Word Error Rate (WER) of 1.8% over Mel Frequency Cepstral Coefficients (MFCCs) and 2.1% over Perceptual Linear Prediction (PLP) features. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Hidden Markov models

来源：评论

学校读者我要写书评

暂无评论

Email Validator Dashboard: A Comprehensive Tool for Accurate Email Validation

Email Validator Dashboard: A Comprehensive Tool for Accurate...

引用

2025 International Conference on Computing for Sustainability and Intelligent Future, COMP-SIF 2025

作者： Chand, Syed Riyyan Ahmed, Uzair Ismail, T Mohammed Kumar, B.P. Pradeep Chand, Syed Bilal Chaitra, P. Design Atria Institute of Technology Dept. of Computer Science Bangalore India Atria Institute of Technology Dept. of Computer Science Engineering Bangalore India HostUp Cloud Technologies Private Limited Bangalore India

ISBN: (纸本)9798331538538

Email authentication is of the utmost importance in maintaining the reliability and quality of email communication, specifically in database management and bulk email marketing. The centerpiece of the project is the development of an Email Validator Dashboard, which is a very user-friendly functionality filled with other useful parts that guarantee the accurate verification of email addresses. The dashboard really presents valuable inputs quickly in terms of deliverability, risk factors, and domain reliability due to its dynamic visualization along with real-time email validation procedures. The project's structure is such that the users get real-time data in the form of interactive charts and graphs, which make them see the transformations occurring as validation is going on, thus distinguishing it from the traditional tools. It comes with cutting-edge back-end data integrity validation processes, such as mailbox verification, domain DNS checks, and syntax verification, alongside powerful analytics. This project is really useful for organizations and developers who work with large email lists and at the same time can ensure high deliverability and compliance with data standards. © 2025 IEEE.

关键词： Marketing

来源：评论

学校读者我要写书评

暂无评论

Unveiling the web of privacy: A comprehensive review of threats to information security in social networks 4th

Unveiling the web of privacy: A comprehensive review of thre...

引用

4th International Conference on Computational Methods in science and technology, ICCMST 2024

作者： Singh, Shalini Baloni, Dev Sarkar, Manash Department of Computer Science and Engineering Quantum University Roorkee India Department of Computer Science and Engineering Atria Institute of Technology Banglore India

ISBN: (纸本)9781032911571

This review paper explores emerging threats to information privacy and security within the dynamic landscape of Online Social Networks (OSNs), which serve as repositories of vast amounts of user data. The rise in social media use emphasizes how crucial it is to inform users about the dangers that come with it, like cyber-attacks and privacy violations. Even though various social media platforms have been made possible by technological advancements, consumers’ ignorance of these platforms creates serious security risks. Although academics have come up with creative security solutions for social networks, there are obstacles in the way of their actual application. It is essential to continuously assess security vulnerabilities in light of technology improvements. This study examines several dangers to online social networks and suggests countermeasures based on a variety of models, frameworks, and encryption methods. Despite advancements, privacy concerns remain unaddressed, requiring parents to actively supervise their children’s online activity. © 2025 the Author(s).

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

Stairdepth: a novel staircase detection through depth maps generated by depth anything V2

引用

International Journal of Information technology (Singapore) 2025年第4期17卷 2075-2086页

作者： Shekar, Avire Laxmi Chandra Chandrika, Mukkolla Bhuvana Naidu, Vakkalagadda Hemanth Muppalaneni, Naresh Babu Department of Computer Science and Engineering Indian Institute of Information Technology Design and Manufacturing Kurnool Kurnool India

Detection of a staircase is an important task in the fields of both assistive technology and autonomous navigation with an aim to enable substantial improvement in safety and accessibility for both those with limited vision and those with mobility problems. Classic methods often perform poorly under challenging conditions, such as low lighting, irregular architectural arrangements, or occluded views. Whereas the integration of physical depth sensors has increased detection accuracy, this integration introduces increased complexity, cost and greater maintenance issues. Our work proposes a novel architecture capable of directly estimating depth maps from color images using a pre-trained model without relying on physical sensors for depth estimation. It detects the staircase efficiently, including the classification of stair lines into concave or convex, with the classification of surfaces as a tread or riser, hence localizing staircases accurately. Our proposed methodology demonstrated exceptional results, with 83.2% Precision, 90.8% Recall, and 72.2% Intersection over Union on stair line classification and 97.1% Pixel Accuracy and 82.9% Mean Pixel Accuracy on stair surfaces by outperforming the efficacy of current approaches pertaining to stair line and surface classification. © Bharati Vidyapeeth's institute of computer Applications and Management 2025.

关键词： Depth image Segmentation Stair surfaces Staircase detection StairDepth Stairline

来源：评论

学校读者我要写书评

暂无评论

Decentralized MIMO Systems With Imperfect CSI Using LMMSE Receivers

引用

IEEE Journal on Selected Topics in Signal Processing 2025年第2期19卷 331-348页

作者： Zhuang, Zeyan Zhang, Xin Xu, Dongfang Song, Shenghui Eldar, Yonina C. The Hong Kong University of Science and Technology Department of Electronic and Computer Engineering Hong Kong Hong Kong The Hong Kong University of Science and Technology Division of Integrative Systems and Design Hong Kong Hong Kong The Weizmann Institute of Science Department of Mathematics and Computer Science Rehovot7610001 Israel

Centralized baseband processing (CBP) is required to achieve the full potential of massive multiple-input multiple-output (MIMO) systems. However, due to the large number of antennas, CBP suffers from two major issues: 1) Extensive data interconnection between radio frequency (RF) circuitry and the central processing unit;and 2) high-dimensional computation. To this end, decentralized baseband processing (DBP) has been proposed, where the antennas at the base station are partitioned into clusters connected to separate RF circuits and equipped with separate computing units. However, the optimal fusion scheme that maximizes signal-to-interference-and-noise ratio (SINR) and the related performance analysis for DBP with general spatial correlation and imperfect channel state information (CSI) have not been studied. In this paper, we consider a decentralized MIMO system where all clusters adopt linear minimum mean-square error (LMMSE) receivers. We first establish an optimal linear fusion scheme that has high computational and data input/output costs. To reduce the cost, we then propose two suboptimal fusion schemes with reduced complexity. For all three schemes, we study the SINR performance by leveraging random matrix theory and demonstrate conditions under which the suboptimal schemes are optimal. Furthermore, we determine the optimal regularization parameter for the LMMSE receiver, identify the best antenna partitioning strategy, and prove that the SINR will decrease as the number of clusters increases. Numerical simulations validate the accuracy of the theoretical results. © 2025 IEEE.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

Driver Drowsiness Detection System

Driver Drowsiness Detection System

引用

Smart Systems for applications in Electrical sciences (ICSSES), International Conference on

作者： Rimsha Mehnaaz Khan B P Pradeep Kumar Sidra Kabir Yash Raj Department of Computer Science and Design Atria Institute of Technology VTU Bengaluru Karnataka India

ISBN: (数字)9798331507671

ISBN: (纸本)9798331507688

Systems to identify tiredness in drivers are becoming more widely recognized as essential safety innovations intended to reduce accidents brought on by fatigued drivers. This paper delivers an inclusive analysis of the technology and approaches used in these systems, which primarily focus on keeping an eye on the driver alertness using a range of techniques, such as behavioural analysis, physiological measures, and computer vision. These systems can effectively detect indicators of tiredness, like head position and eye closure length, by utilizing real-time data from cameras and sensors. In addition to warning drivers when weariness is detected, the installation of such systems enhances overall vehicle safety by integrating with cutting-edge driver aid systems. Considering the startling data regarding sleepy driving, which contributes significantly to traffic accidents.

关键词： computer vision Technological innovation Streaming media Real-time systems Physiology Magnetic heads Facial features Faces Vehicles Accidents

来源：评论

学校读者我要写书评

暂无评论

Adopting Graph Neural Networks to Understand and Reason About Dynamic Driving Scenarios

IEEE Open Journal of Intelligent Transportation Systems

引用

IEEE Open Journal of Intelligent Transportation Systems 2025年 6卷 579-589页

作者： Su, Peng Xiang, Conglei Chen, Dejiu KTH Royal Institute of Technology Faculty of Engineering Design Department Stockholm100 44 Sweden KTH Royal Institute of Technology Electrical Engineering and Computer Science School Stockholm100 44 Sweden

With advances in Deep Neural Networks (DNN), Automated Driving Systems (ADS) enable the vehicle to perceive their surroundings in dynamic driving scenarios and perform behaviors by collecting operational data from sensors such as LiDAR and cameras. Current DNN typically detect objects by analyzing and classifying unstructured data (e.g., image data), providing critical information for ADS planning and decision-making. However, advanced ADS, particularly those required to perform the Dynamic Driving Task (DDT) autonomously, are expected to understand driving scenarios across various Operational design Domains (ODD). This capability requires the support for a continuous comprehension of driving scenarios according to operational data collected by sensors. This paper presents a framework that adopts Graph Neural Networks (GNN) to describe and reason about dynamic driving scenarios via analyzing graph-based data based on collected sensor inputs. We first construct the graph-based data using a meta-path, which defines various interactions among different traffic participants. Next, we propose a design of GNN to support both the classification of the node types of objects and predicting relationships between objects. As results, the performance of the proposed method shows significant improvements compared to the baseline method. Specifically, the accuracy of node classification increases from 0.77 to 0.85, while that of relationships prediction rises from 0.74 to 0.82. To further utilize graph-based data constructed from dynamic driving scenarios, the proposed framework supports reasoning about operational risks by analyzing the observed nodes and relationships in the graph-based data. As a result, the model achieves a MRR of 0.78 in operational risks reasoning. To evaluate the practicality of the proposed framework in real-world systems, we also conduct a real-time performance evaluation by measuring the average process time and the Worst Case Execution Time (WCET). Com

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Universal Readability: Simplification of Complex Text Using Deep Learning

Universal Readability: Simplification of Complex Text Using ...

引用

Smart Systems for applications in Electrical sciences (ICSSES), International Conference on

作者： B. P. Pradeep Kumar Sneha R Raksha R Shrajan Shetty Dept. of Computer Science and Design Atria Institute of Technology Bangalore India

ISBN: (数字)9798331507671

ISBN: (纸本)9798331507688

The proposed framework implements text simplification through Bidirectional Encoder Representations from Transformers (BERT) model structures together with Generative Pre-trained Transformer-2 (GPT-2) deep learning elements. Traditional rule-based and statistical methods cannot accurately achieve the balance between simplification while retaining the original meaning. The evaluation results yielded a Bilingual Evaluation Understudy (BLEU) score of 0.95 while demonstrating superior readability performance. The research evaluation established successful text simplification because the Simple Measure of Gobbledygook (SMOG) reading difficulty decreased while the Flesch Reading Ease increased. The sophisticated technique delivers better outcomes than standard testing systems because it works efficiently with various types of text documents. The method benefits educational institutions and technical tasks with outstanding practicality and solves accessibility requirements effectively. Future research should integrate diverse data with domain-specific features to improve performance of the model.

关键词： Deep learning Statistical analysis Computational modeling Bidirectional control User interfaces Transformers Readability metrics Encoding User experience Testing

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：