检索结果-内蒙古大学图书馆

IEEE Students' Conference on Electrical, Electronics and Computer science (SCEECS)

作者： Angelina George B. M. Beena Department of Computer Science & Engineering Amrita School of Computing Bengaluru Amrita Vishwa Vidyapeetham India

ISBN: (数字)9798331529833

ISBN: (纸本)9798331529840

Data centers are vital to modern computing but contribute significantly to carbon emissions due to their high energy consumption. To address this, machine learning models can optimize resource allocation by analyzing sensor metrics. In this study, the Random Forest Regressor was chosen for its ability to capture sensor behavior, even showing negative R 2 values, aligning with observed patterns. Meanwhile, k-Nearest Neighbors (kNN) emerged as the best model for learning from the same data, despite irregularities, with R 2 values ranging between 0.6 and 1. This analysis supports sustainable data center management by enhancing energy efficiency and minimizing environmental impact. Future work will explore the integration of these models into simulation environments.

关键词： Data centers Analytical models Biological system modeling Computational modeling Nearest neighbor methods Green computing Data models Resource management Sustainable development Random forests

来源：评论

学校读者我要写书评

暂无评论

Digital Guardians: Exploring Video Analytics in Autonomous Classroom Surveillance Systems 2

Digital Guardians: Exploring Video Analytics in Autonomous C...

引用

2nd International Conference on Machine Learning and Autonomous Systems, ICMLAS 2025

作者： Sreekar, K. Jeyakumar, G. Amrita School of Computing Amrita Vishwa Vidyapeetham Department of Computer Science and Engineering Coimbatore India

ISBN: (纸本)9798331505745

The growing prevalence of intelligent surveillance systems has expedited progress in video analytics, particularly in learning environments. This survey article analyzes various methods for assessing student behavior and extracting features from classroom videos, with a focus on identifying test malpractices. We look at many existing systems that use techniques like facial identification, tracking, gesture recognition, and engagement analysis to monitor student activities. Here we shed light on the research needs by showing the benefits and limitations of several strategies that involve coping with occlusions, variable lighting, and small face sections in video frames. Our goal is to develop a strong system that uses live video feeds to detect test malpractices in a better and accurate manner while being efficient. © 2025 IEEE.

关键词： Gesture recognition

来源：评论

学校读者我要写书评

暂无评论

Enhancing Underwater Image Captioning Using Transformer Models and Augmented Terrestrial Datasets

Enhancing Underwater Image Captioning Using Transformer Mode...

引用

Pervasive Computational Technologies (ICPCT), International Conference on

作者： Rohan Gamidi M Hemasri Tejaswi Muppala Vinitha Chowdary A Aiswariya Milan K Suja Palaniswamy Department of Computer Science and Engineering Amrita School of Computing Bengaluru Amrita Vishwa Vidyapeetham India

ISBN: (数字)9798331508685

ISBN: (纸本)9798331519476

Underwater environments present unique challenges for image analysis due to light attenuation, scattering, and color distortion, which significantly degrade image quality. These facts make simple operations such as image captioning, which is crucial in marine sciences, search, and conservation, challenging. Current datasets such as Flickr8K contain mostly terrestrial images and do not possess features distinctive to underwater environments therefore direct training of models is not possible in this context. To cover this gap, the current work combines a Transformer-based image captioning model with an enhanced underwater-specific image augmentation set. These augmentations mimic real-life underwater consequences such as RGB attenuation, Gaussian blur, muddy particles, noise, and gradient-based illumination distortions. The implemented model, learned and recursively tuned on the Flickr8k dataset, proves the ability to generate reasonable captions for underwater images even though the dataset belongs to Flickr8k providing a terrestrial environment. This is proven by the enhanced BLEU scores as well as from the results of the experiment which trained the translation model in less than ten percent of the dataset to show its flexibility on underwater scenarios but without necessitating the use of an underwater dataset.

关键词： Training Computer vision Visualization Accuracy Computational modeling Lighting Transformers Attenuation Distortion Image augmentation

来源：评论

学校读者我要写书评

暂无评论

Zokfuzz: Detection of Web Vulnerabilities via Fuzzing 2

Zokfuzz: Detection of Web Vulnerabilities via Fuzzing

引用

2nd International Conference on Consumer Electronics and Computer engineering, ICCECE 2022

作者： Zhang, Hao Dong, Weiyu Jiang, Liehui School of Cyber Science and Engineering Zhengzhou University Zhengzhou China State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781665408868

Fuzzing is one of the mainstream web application automated vulnerability detection methods. Because of its black box characteristics, it can be used to detect vulnerabilities without knowing the source code of the target application, so it is widely used. However, traditional web application fuzzing methods have the disadvantages of limited test cases, slow execution process, and low efficiency. In order to solve these problems, some scholars introduced genetic algorithms in the fuzzing process. However, after the test cases are entered, it is difficult to automatically detect the state of the web application, which will lead to deviations in the evolution direction of the population and the high aggressiveness of the test cases. This paper proposes a web application fuzzing script, which can realize automatic monitoring of web application status changes, and at the same time more accurately judge whether the test case is offensive. On this basis, an improved mutation method is proposed to enable the population to quickly evolve more aggressive test cases. Experimental results show that ZokFuzz can detect more web vulnerabilities compared to X-Ray and Burp Suite. © 2022 IEEE.

关键词： Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

Extracting Triplets from Domain-specific Texts using Fine-Tuned LLaMA Models: A Comparative Study

Extracting Triplets from Domain-specific Texts using Fine-Tu...

引用

3rd International Conference on Machine Learning and Data engineering, ICMLDE 2024

作者： Veena, G. Rajendran, Aparna Gupta, Deepa Department of Computer Science and Applications Amrita School of Computing Amrita Vishwa Vidyapeetham Amritapuri India Dept of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bengaluru India

This research centers on extracting triplets (Subject, Verb, Object) from domain-specific texts, particularly focusing on the film industry. We compiled a corpus of 4,300 sentences from 500 well-known movie articles and Wikipedia entries. The study targeted five key relations: Composed by, Written by, Released on, Made f rom, and Released in, along with their respective entities. Dependency parsing was employed to extract triplets from this dataset. These extracted triplets were then used for supervised fine- tuning of the LLaMa model. We assessed the evaluation loss across different versions of the LLaMA models and found that LLaMA 3 outperformed the others, showing the best performance for this task. For testing, the fine-tuned LLaMA model is used to extract triplets from unseen sentences. The extracted triplets can be used to create knowledge graphs, offering structured representations of information by mapping relationships between entities. © 2024 The Authors. Published by ELSEVIER B.V.

关键词： Finetuning LLaMA LLM NLP Zero-shot

来源：评论

学校读者我要写书评

暂无评论

Improving the review classification of Google apps using combined feature embedding and deep convolutional neural network model

引用

Journal of Ambient Intelligence and Humanized computing 2023年第4期14卷 4257-4272页

作者： Aslam, Naila Alzamzami, Ohoud Xia, Kewen Sadiq, Saima Umer, Muhammad Bisogni, Carmen Ashraf, Imran School of Electronics and Information Engineering Hebei University of Technology Tianjin China Faculty of Computing and Information Technology King Abdulaziz University Jeddah Saudi Arabia Department of Computer Science Khwaja Fareed University of Engineering and Information Technology Rahim Yar Khan Pakistan Department of Computer Science Information Technology The Islamia University of Bahawalpur Bahawalpur Pakistan Department of Computer Science University of Salerno Fisciano Italy Information and Communication Engineering Yeungnam University Gyeongsan Korea Republic of

Online reviews play an integral part in making mobile applications stand out from the large number of applications available on the Google Play store. Predominantly, users consider posted reviews for appropriate app selection. Manual categorization of such reviews is both inefficient and time-consuming. Therefore, automatic analysis of the sentiments of such reviews provides fast suggestions for new users and facilitates their selection of the appropriate app. However, data imbalance is a major challenge for performing class prediction of such reviews as their distribution is sparse and often leads to low accuracy. This work proposes a framework to overcome this limitation. Extensive experiments are performed using the original and balanced data with the synthetic minority oversampling technique (SMOTE) and adaptive synthetic sampling (ADASYN). Additionally, deep learning and machine learning models are evaluated using FastText, FastText Subword, global vector (GloVe), and their combinations for word representation. Baseline machine learning models, including random forest, extra tree classifier, gradient boosting, Naive Bayes, logistic regression (LR), stochastic gradient descent (SGD), and voting classifier (VC) that combines LR and SGD, are used for comparison. The outcomes show that the convolutional neural network using a combination of word embedding techniques produces the most accurate results. © 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

ArtLens: User-Driven Customization in AI-Powered Art Generation with Blip2 and Stable Diffusion 6

ArtLens: User-Driven Customization in AI-Powered Art Generat...

引用

6th International Conference on Futuristic Trends in Networks and computing Technologies, FTNCT 2024

作者： Ranjit, Anargha Gaadha, V.S. Sreerag, Aswin Bonthu, Yasaswini Geetha, M. Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Amritapuri India

AI is changing the interpretation and teaching of art in the digital humanities with models like Blip2 and Stable Diffusion. This paper evaluates these models but particularly looks at how Blip2 can integrate visual and textual data to caption artworks, classify styles, interpret themes, and conduct contextual analysis, while Stable Diffusion can generate high-resolution images from literary prompts. The proposed method involves the use of Blip2 in creating descriptive captions for artworks and then modified with zero-shot prompting based on specific user demands. This updated text combined with the original input is then fed into the Stable Diffusion model to create new visuals reflecting the changed aspects. This method permits greater flexibility and user-driven customisation in creative image production. This work is a demonstration of the use of AI in interpretation with dynamic and interactive art and its ability to provide user-oriented, more subtle and personalized outputs. The embedding of user-specific adjustments into the workflow has enhanced the creative expression and adaptability of the generated content by AI, offering new avenues of applications in digital humanities, art education, and curatorial practices. This study marks the revolutionary potential of AI in most sectors, reflecting how it may inspire new involvement with art and culture. © 2025 The Author(s). Published by Elsevier B.V.

关键词： User centered design

来源：评论

学校读者我要写书评

暂无评论

An Empirical Comparative Study on Pruning and Quantization Algorithms for Model Compression 2

An Empirical Comparative Study on Pruning and Quantization A...

引用

2nd International Conference on Machine Learning and Autonomous Systems, ICMLAS 2025

作者： Mukesh, Narravula Kumar, Mallisetty Rathna Vikas, Ambati Sai Jeyakumar, Gurusamy Amrita School of Computing Coimbatore Amrita Vishwa Vidyapeetham Department of Computer Science and Engineering India

ISBN: (纸本)9798331505745

In a world where Neural networks and Deep learning have become of major importance, the need for methods to reduce their memory storage and usage has become of major importance. Hence the need for quantization and pruning has become huge and important. Also, the hardware used also decides the necessity of memory reduction due to the hardware's compatibility and memory computation constraints. This study compares different pruning and quantization models and reports about which are the best methods among the different methods considered in this study. The proposed comparison is carried out using the metrics of prediction accuracy and the time taken for training. This study identifies that magnitude-based pruning is the best among all the CPU-based systems with an accuracy of 0.998, whereas structured pruning has the highest accuracy of 0.985. The focus is on the hardware used, which are mainly the GPU-based systems and the CPU-based systems. The study reveals that the pruning method is better in the faceoff between pruning and quantization methods in terms of accuracy and computational time. When compared between the CPU and the GPU, concerning accuracy CPU holds the edge, whereas concerning the computational time the GPU holds the edge. Overall, concerning the balance between accuracy and computational time, the GPU-based systems hold the upper hand, making them more suitable for the optimization task. © 2025 IEEE.

关键词： Central Processing Unit Graphical Processing Unit Neural Networks Pruning Quantization

来源：评论

学校读者我要写书评

暂无评论

Scheduling Policies in Edge computing for Real-Time Video Analytics

Scheduling Policies in Edge Computing for Real-Time Video An...

引用

Intelligent Data Communication Technologies and Internet of Things (IDCIoT), International Conference on

作者： Shreevatsan S S Yasasree Lasya Annambhotla Anaswara Reji Shinu M Rajagopal Department of Computer Science and Engineering Amrita School of Computing Bengaluru Amrita Vishwa Vidyapeetham India

ISBN: (数字)9798331527549

ISBN: (纸本)9798331527556

This study compares different scheduling algorithms used in edge computing for real-time video processing. Instead of sending video data to cloud servers, the video is processed directly at the edge, reducing delays and saving bandwidth. The focus is on how different scheduling methods manage resources, prioritize tasks, and reduce the time it takes to analyze video. The algorithms are tested in simulated environments to understand how they impact system efficiency and performance. The goal of this work is to find the best scheduling policies for real-time video processing at the edge, improving overall system performance and scalability.

关键词： Scheduling algorithms Visual analytics System performance Scalability Streaming media Load management Real-time systems Servers Edge computing Testing

来源：评论

学校读者我要写书评

暂无评论

Post Quantum Blockchain Based Lightweight Edge Server Handover Authentication Key Agreement Protocol for Vehicular Communication System 14

Post Quantum Blockchain Based Lightweight Edge Server Handov...

引用

14th International Conference on computing Communication and Networking Technologies, ICCCNT 2023

作者： Sharma, Tannu Durgarao, M.S.P. Reddy, Kethamreddy Karthikeya Kaushal, Sachin Sreedevi, A.G. Chaudhary, Dharminder Lovely Professional University Dept. of Mathematics Jalandhar India Amrita Vishwa Vidyapeetham Amrita School of Computing Dept. of Computer Science & Engineering Chennai India

ISBN: (纸本)9798350335095

We have designed a safe and portable vehicle to infrastructure handover authentication using blockchain technology. In this protocol, blockchain technology provides benefits for data sharing since it can guarantee data quality, transparency, and provenance. To get information for handover authentication, road side units can act nodes of the consortium blockchain. Furthermore, transactions on the blockchain cannot be falsified since they include an road side unit signature and cannot be updated due to blockchain technological features. As a result, other road side units may verify the authenticity of vehicles by utilizing the data stored on the blockchain. For authentication phase, we have used lattice based cryptography that plays a very important role in authentication and key exchanges that protects against the threat of quantum attacks. The proposed protocol ensures both securities against quantum attacks and efficiency due to simple algebraic operations that are polynomial addition and multiplications. © 2023 IEEE.

关键词： Blockchain

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：