检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Wang, Chengyu Zhang, Taolin Hong, Richang Huang, Jun Alibaba Cloud Computing School of Computer Science and Information Engineering Hefei University of Technology China

Recently, the reasoning capabilities of large reasoning models (LRMs), such as DeepSeek-R1, have seen significant advancements through the slow thinking process. Despite these achievements, the substantial computational demands of LRMs present considerable challenges. In contrast, small reasoning models (SRMs), often distilled from larger ones, offer greater efficiency and can exhibit distinct capabilities and cognitive trajectories compared to LRMs. This work surveys around 170 recently published papers on SRMs for tackling various complex reasoning tasks. We review the current landscape of SRMs and analyze diverse training and inference techniques related to SRMs. Furthermore, we provide a comprehensive review of SRMs for domain-specific applications and discuss possible future research directions. This survey serves as an essential reference for researchers to leverage or develop SRMs for advanced reasoning functionalities with high efficiency. Copyright © 2025, The Authors. All rights reserved.

关键词： Surveying

来源：评论

学校读者我要写书评

暂无评论

Predicting Student Success in Online Learning Using Machine Learning and Explainable AI 3

Predicting Student Success in Online Learning Using Machine ...

引用

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

作者： Gayatri, Phani S. Dinesh, Gopika Gupta, Deepa Nair, Aarathi Rajagopalan Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bengaluru India

ISBN: (纸本)9798331543891

With the quick expansion of online education, understanding user engagement and performance metrics has become crucial for improving learning outcomes. This study analyzes a dataset which encompasses comprehensive details about user demographics, engagement levels and course performance in online learning environments using machine learning. The motivation behind this work is to uncover factors influencing course completion rates, engagement patterns and overall learning success by exploring eleven different ML models along with the comparison of three data balancing/sampling and six feature selection techniques. The best of the four data balancing techniques, TOMEK Links and the best of the six feature selection techniques, Chi-square test are preferred based on the model performances examined using evaluation metrics and the time taken by the models. Out of all the models, the five best performing models are LightGBM, XGBoost, Decision Tree (entropy), KNN and ADA Boost. Finally, the explainable AI tools LIME and SHAP are used for model interpretation. Decision Tree with accuracy 1 and highest macro F1 score 0.97 became the best model which executed within 0.37 seconds being the least time. © 2025 IEEE.

关键词： Curricula

来源：评论

学校读者我要写书评

暂无评论

Efficient Quality Controllable Neural Image Compression based on QD-Model

Efficient Quality Controllable Neural Image Compression base...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Shaokang Wang Guoqing Xiang Jinchang Xu Shanghang Zhang Xiaodong Xie School of Computer Science Peking University Beijing China Alibaba Cloud Computing Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Neural image compression has achieved significant advancements, consistently outperforming traditional codecs in terms of performance. However, research on quality control algorithms for neural image compression is still lacking. In this paper, we propose a framework designed to control the quality of compressed images through a one-pass pre-analysis. First, we construct a foundational relationship between the quantization factor and compression distortion, utilizing variable rate neural image compression as the basis for quality control. Second, we introduce the image Content-Compression features-based Distortion Estimation Network (C 2 DEN) to efficiently fit the sample-adaptive Quantization-Distortion (QD) model. Leveraging the QD model, we convert the target quality into a quantization factor to control the compression model, enabling quality-controllable compression of samples. Experimental results show that the average quality errors on four different datasets are only 0.89%, 1.79%, 1.73%, and 1.61%. Compared with existing control methods, our method reduces the algorithm time complexity by 98.58%, 98.52%, 98.85%, and 98.50% while ensuring accuracy, which further demonstrates the superiority of our method.

关键词： Image coding Quantization (signal) Acoustic distortion Signal processing algorithms Rate-distortion Estimation Quality control Rate distortion theory Time complexity Speech processing

来源：评论

学校读者我要写书评

暂无评论

Classification in Viva Interactions: Distinguishing Questions, Answers, and Statements with BERT Embedding and SVM

Classification in Viva Interactions: Distinguishing Question...

引用

Knowledge Engineering and Communication Systems (ICKECS), International Conference on

作者： Y.S.S Rohit Tangudu Rushik V. Hyndhava Peeta Basa Pati Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bengaluru India

ISBN: (数字)9798331537012

ISBN: (纸本)9798331537029

While oral vivas are important in judging student content mastery and presentation approaches, their manual evaluation can be quite difficult due to interfered speech components, intonation differences and the need to differentiate between different types of dialogue. For these challenges, we introduce a model that incorporates machine learning (ML) and natural language processing (NLP) technologies to automatically segment and label viva recordings into the components: question, answer, and statement. The system achieves its best results using SVM with BERT embedding F1-score of 0.81, ensuring accurate segmentation and reliable performance. In exploiting oral and prosodic features, this system improves segmentation precision, reduces transcription time, and enables evaluators to focus on content evaluation. This technology facilitates objective evaluation in ICT(Information communication technology) enfacilitated learning and it is a useful, affordable learning tool for teachers as well as students.

关键词： Support vector machines Vocabulary Accuracy Scalability Stacking Semantics Transformers Question answering (information retrieval) Reliability Speech processing

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Framework for Comparative Analysis in Flood Detection and Area Estimation

A Deep Learning Framework for Comparative Analysis in Flood ...

引用

Interdisciplinary Approaches in Technology and Management for Social Innovation (IATMSI), IEEE Conference on

作者： Shiva Teja Pecheti Ashwaj U Uthej K Murali K Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bengaluru India

ISBN: (数字)9798331521691

ISBN: (纸本)9798331521707

This research presents an innovative deep learning framework designed to be utilized for area estimation and flood detection. The most suitable architecture for real-time applications is determined using this framework by a thorough comparison of advanced architectures, including U-Net, LeNet, ResNet, and VGG16. A crucial initial phase is to classify images as flooded or non-flooded using an effective Multilayer Perceptron (MLP) classifier. Users can see areas damaged by flooding and get quick assessments with the help of a simple Streamlit interface suited to foster immediate engagement. More significantly, the results show that LeNet is the most reliable model all things considered, often estimating flooded areas with the highest accuracy. Additionally, the framework delivers essential data for rapid disaster response by estimating the intensity of floods through accurate pixel analysis. The study offers helpful techniques to improve planning and response strategies for floods in an era of growing environmental uncertainty, as well as to giving important new insights into the use of deep learning in the management of floods.

关键词： Deep learning Technological innovation Accuracy Uncertainty Estimation Streaming media Real-time systems Floods Reliability Testing

来源：评论

学校读者我要写书评

暂无评论

Enhancing Agricultural Advisory Services with Multilingual LLaMA and RAG

Enhancing Agricultural Advisory Services with Multilingual L...

引用

Next Generation Communication & Information Processing (INCIP), International Conference on

作者： G. Bharathi Mohan C.M. Jayanth Adhitya A. Mithilesh Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Chennai India

ISBN: (数字)9798331528140

ISBN: (纸本)9798331528157

Agriculture, being a vital component of global food security, necessitates the development of innovative solutions to address the knowledge gap experienced by numerous farmers, particularly in developing countries where access to expert advice is scarce. In these areas, agricultural producers often depend on helplines for essential guidance, but the exorbitant costs and limited accessibility of these services create substantial obstacles. By automating responses to agricultural inquiries, the burden on traditional helpline systems can be alleviated, enabling farmers to access prompt and precise information. The combination of artificial intelligence and agriculture offers a chance to tackle these challenges, with advanced language models, especially transformers, demonstrating significant potential in comprehending intricate agricultural queries and delivering appropriate responses. This paper investigates how large language models (llms) can simplify the process of finding answers for farmers by utilizing their advanced language processing abilities. By analyzing a vast collection of over four million farmer inquiries from tamil nadu, india, encompassing diverse agricultural scenarios, this study showcases the efficacy of llms in bridging information gaps and equipping farmers with immediate access to crucial knowledge.

关键词： Precision agriculture Large language models Biological system modeling Food security Information processing Metadata Transformers Solids Natural language processing Multilingual

来源：评论

学校读者我要写书评

暂无评论

SmartKitch: AI-Powered Kitchen Inventory Management and Personalized Culinary Assistant

SmartKitch: AI-Powered Kitchen Inventory Management and Pers...

引用

Knowledge Engineering and Communication Systems (ICKECS), International Conference on

作者： A.R Dhedeep Reddy Shamil Saidu Mohamed Cj Harini S Ullas Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bangalore India

ISBN: (数字)9798331537012

ISBN: (纸本)9798331537029

SmartKitch is an intelligent kitchen management system designed to enhance efficiency in meal planning and grocery management. It provides real-time recipe recommendations based on available inventory and allows users to track and manage their kitchen supplies. The system generates shopping lists, offers personalized dietary advice, and integrates a com- prehensive recipe repository powered by ChatGPT for detailed culinary guidance. SmartKitch supports data export/import, advanced search and filtering options, and user authentication features. To ensure secure user access, it employs an encryption algorithm for authentication. Additionally, user data is securely stored using encrypted databases to maintain privacy. The system seamlessly integrates with an SQLite database, enabling efficient data management and synchronization.

关键词： Surveys computer vision Databases Authentication Chatbots Planning Synchronization Reliability Sustainable development Testing

来源：评论

学校读者我要写书评

暂无评论

Automated Web Application Security Reporting System with Slack Integration

Automated Web Application Security Reporting System with Sla...

引用

Advancements in Electrical, Electronics, Communication, computing and Automation (ICAECA), International Conference on

作者： M. L. Sahasra Reddy Raja Varshini S. Saravanan Department of Computer Science and Engineering Amrita School of Computing Chennai Amrita Vishwa Vidyapeetham India

ISBN: (数字)9798331525439

ISBN: (纸本)9798331525446

Ensuring web application security is no longer an option but a basic need in today's connected world. The disturbing rate of cyber threats has made the assurance of secure web applications crucial today. It is important to find vulnerabilities in the initial development stage of the web application. Therefore, this work proposes a new approach that integrates vulnerability scanning natively within Slack, a messaging platform for application development team collaboration, that makes security procedures easier to carry out by the developers. By making use of a Slack slash command, a developer can upload a Uniform Resource Locator (URL) of a web application with the request to have a scan of vulnerabilities, such as Structured Query Language (SQL) Injection, Cross-Site Scripting, Session Hijacking, Sensitive Data Leakage, Broken Authentication, Command Injection, Cross-Origin Resource Sharing (CORS) misconfiguration, Cache Poisoning, Clickjacking, Cross-Site Request Forgery, Insecure Direct Object References (IDOR), Insecure APIs, Lightweight Directory Access Protocol (LDAP) Injection, Path Traversal and Subdomain Takeover. The proposed approach uses the Google Gemini API to generate precise reports on each vulnerability based on potential risks as well as suggested mitigation strategies, enabling them to rapidly determine and conquer security flaws. The solution equips developers with the ability to proactively secure applications using a defense-first approach in modern development practice. The proposed approach contributes to ever changing field of automated security solutions in a user-prioritized manner.

关键词： Uniform resource locators Structured Query Language Prevention and mitigation SQL injection Session hijacking Application security Telecommunication computing Internet Security Resource management

来源：评论

学校读者我要写书评

暂无评论

REMOTELY CONTROLLABLE DNA MICROFLOW 26

REMOTELY CONTROLLABLE DNA MICROFLOW

引用

26th International Conference on Miniaturized Systems for Chemistry and Life sciences, MicroTAS 2022

作者： Udono, Hirotake Nomura, Shin-Ichiro M. Takinoue, Masahiro Department of Computer Science School of Computing Tokyo Institute of Technology Japan Department of Robotics Graduate School of Engineering Tohoku University Japan

ISBN: (纸本)9781733419048

Despite an emerging interest in fluid manipulation within confined environment such as biological cells and micro-scale biomimics, remotely controllable micro-scale flow remains unexplored. However, intracellular microfluidics is not compatible with established top-down microfluidic methods because of limited accessibility. Here, we demonstrate photoresponsive DNA fluid, condensate of DNA nanostructures (DNA motifs), as a versatile mechanistic tool for the remote control of microflow. Insertion site of azobenzene, introduced for photocontrollability, determines an extent of dynamic shift in binding enthalpy between DNA motifs, providing the underpinning mechanism for drastic and facile flow-mode switching. © 2022 MicroTAS 2022 - 26th International Conference on Miniaturized Systems for Chemistry and Life sciences. All rights reserved.

关键词： Remote control

来源：评论

学校读者我要写书评

暂无评论

Exploring Mental Health Status Correlations Through Topic Modeling on Social Media Data

Exploring Mental Health Status Correlations Through Topic Mo...

引用

Pervasive computing and Social Networking (ICPCSN), International Conference on

作者： Krishna Nanda S Santhanalakshmi Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bengaluru India

ISBN: (数字)9798331535193

ISBN: (纸本)9798331535209

Mental health, once a stigmatized and overlooked topic, is now widely discussed. As concerns related to mental health began to grow, people began to share their personal experiences on various platforms and also began to seek the help of professionals. With this growth, social networks have become a prominent space where people share what they are going through. The main goal of this study is to analyze social media data to uncover and understand key topics linked to different mental health disorders using topic modeling techniques. This could help mental health professionals gain deeper insights into each disorder and it can also help them understand the correlation between usage of language and various mental health disorders. As early identification and intervention are crucial in providing effective treatment and improved long-term results, this work could help identify those topics that contribute to these issues in social media texts. The social media platform is where people express their emotions in the most open and unfiltered way due to anonymity and the benefit of not being under the pressure of in person judgment. This work analyzed over 1,00,000 statements classified into seven mental health disorders such as personality disorders, suicide, stress, anxiety, bipolar, depression, and normal to identify prominent topics related to each. Topic modeling models such as LDA, NMF, Top2Vec, and BERTopic are used to implement this work and identify topics that could reflect a particular mental disorder and also differentiate between disorders. The evaluation is performed based on the factors of coherence score, silhouette score, Calinski-Harabasz score, and human interpretability.

关键词： Pervasive computing Analytical models Correlation Social networking (online) Computational modeling Anxiety disorders Mental health Coherence Depression Data models

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：