检索结果-内蒙古大学图书馆

41st International Conference on Machine Learning, ICML 2024

作者： Pan, Xiaokang Li, Xingyu Liu, Jin Sun, Tao Sun, Kai Chen, Lixing Qu, Zhe Hunan Provincial Key Lab on Bioinformatics School of Computer Science and Engineering Central South University Changsha China Department of Computer Science Tulane University New Orleans United States College of Computer National University of Defense Technology Changsha China School of Computer Science and Technology Xi'an Jiaotong University Xi'an China Institute of Cyber Science and Engineering Shanghai Jiao Tong University Shanghai China

STOchastic Recursive Momentum (STORM)based algorithms have been widely developed to solve one to K-level (K ≥ 3) stochastic optimization problems. Specifically, they use estimators to mitigate the biased gradient issue and achieve near-optimal convergence results. However, there is relatively little work on understanding their generalization performance, particularly evident during the transition from one to K-level optimization contexts. This paper provides a comprehensive generalization analysis of three representative STORM-based algorithms: STORM, COVER, and SVMR, for one, two, and K-level stochastic optimizations under both convex and strongly convex settings based on algorithmic stability. Firstly, we define stability for K-level optimizations and link it to generalization. Then, we detail the stability results for three prominent STORM-based algorithms. Finally, we derive their excess risk bounds by balancing stability results with optimization errors. Our theoretical results provide strong evidence to complete STORM-based algorithms: (1) Each estimator may decrease their stability due to variance with its estimation target. (2) Every additional level might escalate the generalization error, influenced by the stability and the variance between its cumulative stochastic gradient and the true gradient. (3) Increasing the batch size for the initial computation of estimators presents a favorable trade-off, enhancing the generalization performance. Copyright 2024 by the author(s)

关键词： Optimization algorithms

来源：评论

学校读者我要写书评

暂无评论

PAIR: Protein-aptamer interaction prediction based on language models and contrastive learning framework

PAIR: Protein-aptamer interaction prediction based on langua...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zhang, Jun Yan, Zhiqiang Zeng, Hao Zhu, Zexuan Shenzhen University National Engineering Laboratory for Big Data System Computing Technology Shenzhen China Shenzhen University College of Computer Science and Software Engineering Shenzhen China

ISBN: (纸本)9798350386226

Aptamers are single-stranded DNA or RNA oligonucleotides that selectively bind to specific targets, making them valuable for drug design and diagnostic applications. Identifying the interactions between aptamers and target proteins is crucial for these applications. The systematic evolution of ligands by exponential enrichment process, traditionally used for this purpose, is challenging and time-consuming. The resulting aptamers often suffer from limitations in stability and diversity. Computational approaches have shown promise in aiding the discovery of high-performance aptamers, but existing methods are usually constrained by insufficient training data and limited generalizability. Recently, advancements in pre-training large language models have offered a new avenue to mitigate the dependency on large datasets. In this study, we propose a novel method to predict aptamer-protein interactions using large language models within a contrastive learning framework. Experimental results demonstrate that our method exhibits superior generalization and outperforms existing approaches. This method holds promise as a powerful tool for predicting aptamer-protein interactions. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Recycling as a Service: IoT Enabled Smart Waste Management System with Machine Learning

Recycling as a Service: IoT Enabled Smart Waste Management S...

引用

2024 International Conference on Knowledge engineering and Communication Systems, ICKECS 2024

作者： Poonkuzhali, R. Karamath, M. Sugumaran, P. Tharun, M. Arun, S. K. S. Rangasamy College of Technology Department of Information Technology Tiruchengode637215 India Sri Sai Ram Engineering College Computer Science and Business Systems Chennai600044 India

ISBN: (纸本)9798350359688

This paper introduces Smart Waste Management (SWM), a system integrating IoT technology and ML algorithms to enhance waste management efficiency. By leveraging real-time data from IoT sensors in waste bins, SWM dynamically optimizes collection routes and schedules. ML algorithms analyze data on fill levels and environmental factors to predict waste generation patterns, enabling informed decision-making. This adaptive approach minimizes fuel consumption, reduces costs, and lowers carbon emissions. SWM also offers actionable insights through customizable dashboards, empowering authorities to implement proactive measures such as predictive maintenance and targeted public awareness campaigns. Overall, the integration of IoT and ML technologies presents a transformative solution to optimize resource allocation, promote sustainability, and foster a cleaner urban environment. © 2024 IEEE.

关键词： Predictive maintenance

来源：评论

学校读者我要写书评

暂无评论

Big Data and Machine Learning: Barriers and Methods 4

Big Data and Machine Learning: Barriers and Methods

引用

4th International Conference on Innovative Practices in technology and Management, ICIPTM 2024

作者： Agrawal, Monika Goel, Swati William, P. Agwan, Ritesh Bhangale, Sonal C. Ghige, Onkar Kopargaon Sppu Sanjivani College of Engineering Department of Computer Engineering MH Pune India Inderprastha Engineering College Department of Computer Science and Engineering UP Ghaziabad India Kopargaon Sppu Sanjivani College of Engineering Department of Information Technology MH Pune India Kopargaon Sppu Sanjivani College of Engineering Department of Mechatronics MH Pune India Kopargaon Sppu Sanjivani College of Engineering Department of Electrical Engineering MH Pune India

ISBN: (纸本)9798350307757

In this study, the researchers investigate the mutually beneficial relationship that exists between machine learning (ML) and big data, highlighting both the benefits and the challenges that are brought about by the combination of these two domains. Large datasets provide a difficult environment for machine learning (ML) algorithms to extract meaningful insights from, but ML also provides sophisticated tools to assist in navigating through the complexity of these datasets. Even while there is the potential for synergy, their full-fledged adoption is delayed by significant constraints such as the quality of the data, the need for computing, concerns around privacy, and the skills gap. In addition to highlighting recent advancements in algorithmic efficiency, data preparation, and ethical frameworks, this paper investigates the strategic strategies and techniques that are used to address these issues. An in-depth investigation of machine learning models on a variety of datasets is presented, which offers insights into the efficacy of these models and their applicability for instances that occur in the real world, particularly those that involve the pharmaceutical industry. The research sheds light on the ways in which the environment is undergoing transformations and the ways in which practitioners are actively searching for and putting these challenges into practice in order to fully grasp the revolutionary potential of big data and machine learning. © 2024 IEEE.

关键词： Large datasets

来源：评论

学校读者我要写书评

暂无评论

Reparameterized Importance Sampling for Robust Variational Bayesian Neural Networks 41

Reparameterized Importance Sampling for Robust Variational B...

引用

41st International Conference on Machine Learning, ICML 2024

作者： Long, Yunfei Tian, Zilin Zhang, Liguo Xu, Huosheng College of Computer Science and Technology Harbin Engineering University Heilongjiang Harbin China Modeling and Emulation in E-Government National Engineering Laboratory Harbin Engineering University Heilongjiang Harbin China

Mean-field variational inference (MFVI) methods provide computationally cheap approximations to the posterior of Bayesian Neural Networks (BNNs) when compared to alternatives like MCMC. However, applying MFVI to BNNs encounters limitations due to the Monte Carlo sampling problem. This problem stems from two main issues. First, most samples do not accurately represent the most probable weights. Second, random sampling from variational distributions introduces high variance in gradient estimates, which can hinder the optimization process, leading to slow convergence or even failure. In this paper, we introduce a novel sampling method called Reparameterized Importance Sampling (RIS) to estimate the first moment in neural networks, reducing variance during feed-forward propagation. We begin by analyzing the generalized form of the optimal proposal distribution and presenting an inexpensive approximation. Next, we describe the sampling process from the proposal distribution as a transformation that combines exogenous randomness with the variational parameters. Our experimental results demonstrate the effectiveness of the proposed RIS method in three critical aspects: improved convergence, enhanced predictive performance, and successful uncertainty estimation for out-of-distribution data. Copyright 2024 by the author(s)

关键词： Importance sampling

来源：评论

学校读者我要写书评

暂无评论

Optimization and multi-functional predictive performance of advanced superhydrophobic cementitious materials for water-resistant infrastructure

引用

Asian Journal of Civil engineering 2025年第5期26卷 2023-2036页

作者： Upadhye, Sachin Bagde, Prafulla Sange, Sanjay R. Rokade, Achal Chandak, Mayuri A. Patil, Tejas R. Kakade, Niteen T. Shelke, Nilesh Department of Computer Science and Application School of Computer Science and Engineering Ramdeobaba University Maharashtra Nagpur 440013 India Department of Mathematics Shri Ramdeobaba College of Engineering and Management Ramdeobaba University Maharashtra Nagpur 440013 India Shah And Anchor Kutchhi Engineering College Mumbai University Maharashtra Mumbai 400088 India Department of Civil Engineering Priyadarshini College of Engineering Maharashtra Nagpur 440019 India Department of Civil Engineering Priyadarshini Bhagwati College of Engineering Maharashtra Nagpur 440024 India Department of Mechanical Engineering Tulsiramji Gaikwad Patil College of Engineering and Technology Maharashtra Nagpur 441108 India Symbiosis Institute of Technology Nagpur Campus Symbiosis International (Deemed) University Maharashtra Pune 440008 India

The development of water-resistant infrastructure faces critical challenges due to water ingress, which compromises durability and structural integrity levels. Traditional approaches to hydrophobic concrete often fail to address long-term performance under dynamic environmental conditions, with limitations in durability, scalability, and multi-functionality. To overcome these shortcomings, this study introduces advanced superhydrophobic cementitious materials integrating innovative methodologies that simultaneously enhance hydrophobicity, durability, and functionality. Functionalized nanoparticles with long-chain alkyl silanes offer dual benefits of superior hydrophobicity (contact angle > 150°) and enhanced mechanical strength (> 50 MPa), addressing the challenge of balancing water repellency with structural properties. Gradient porosity via 3D printing enables hierarchical hydrophobic textures, reducing capillary absorption by 95% and extending durability in wet-dry cycles. Stimuli-responsive surface modifications embed microcapsules with hydrophobic agents that ensure self-healing, achieving up to 80% recovery in water repellency post-damage. Machine learning optimizes mix formulations, reducing experimental iterations by 70% while ensuring predictive accuracy (> 95%) for targeted performance. Multi-functional composites incorporating piezoelectric nanoparticles harvest ~ 10 µW/cm² energy while maintaining hydrophobicity, demonstrating potential for smart infrastructure applications. Freeze–thaw resistance is enhanced through cryo-protective agents, achieving > 300 cycles without degradation. The proposed methodologies validate their efficacy through a real-world application testing framework, ensuring scalability with minimal loss of water repellency (> 90%) and compressive strength retention (> 85%). This work not only offers groundbreaking solutions for water-resistant infrastructure but also paves the way for sustainable, resilient, and smart const

关键词： Gradient porosity Hydrophobic nanoparticles Self-healing cement Smart infrastructure Superhydrophobic concrete

来源：评论

学校读者我要写书评

暂无评论

CHART:Intelligent Crime Hotspot Detection and Real-Time Tracking Using Machine Learning

引用

computers, Materials & Continua 2024年第12期81卷 4171-4194页

作者： Rashid Ahmad Asif Nawaz Ghulam Mustafa Tariq Ali Mehdi Tlija Mohammed AEl-Meligy Zohair Ahmed University Institute of Information Technology PMAS Arid Agriculture UniversityRawalpindi46000Pakistan Industrial Engineering Department College of EngineeringKing Saud UniversityRiyadh11421Saudi Arabia Jadara University Research Center Jadara UniversityIrbid21110Jordan Applied Science Research Center Applied Science Private UniversityAmman11931Jordan School of Computer Science and Engineering Central South UniversityChangsha410083China

Crime hotspot detection is essential for law enforcement agencies to allocate resources effectively,predict potential criminal activities,and ensure public *** methods of crime analysis often rely on manual,time-consuming processes that may overlook intricate patterns and correlations within the *** some existing machine learning models have improved the efficiency and accuracy of crime prediction,they often face limitations such as overfitting,imbalanced datasets,and inadequate handling of spatiotemporal *** research proposes an advanced machine learning framework,CHART(Crime Hotspot Analysis and Real-time Tracking),designed to overcome these *** proposed methodology begins with comprehensive data collection from the police *** dataset includes detailed attributes such as crime type,location,time and demographic *** key steps in the proposed framework include:Data Preprocessing,Feature engineering that leveraging domain-specific knowledge to extract and transform relevant *** Map Generation that employs Kernel Density Estimation(KDE)to create visual representations of crime density,highlighting hotspots through smooth data point distributions and Hotspot Detection based on Random Forest-based to predict crime likelihood in various *** Experimental evaluation demonstrated that CHART shows superior performance over benchmark methods,significantly improving crime detection accuracy by getting 95.24%for crime detection-I(CD-I),96.12%for crime detection-II(CD-II)and 94.68%for crime detection-III(CD-III),*** designing the application with integrating sophisticated preprocessing techniques,balanced data representation,and advanced feature engineering,the proposed model provides a reliable and practical tool for real-world crime *** of crime hotspots enables law enforcement agencies to strategize effectively,focusing resources on high-risk areas and thereby enhanc

关键词： Crime hotspot heat map kernel density estimation(KDE) support vector machine(SVM)

来源：评论

学校读者我要写书评

暂无评论

An Innovative Method for Fake News Classification using LSTM-RF Approach 4

An Innovative Method for Fake News Classification using LSTM...

引用

4th International Conference on Electronics and Sustainable Communication Systems, ICESC 2023

作者： Yadav, Prashant S Subba Reddy, K. Ashwini, S. Santhosh, J. Nirmala Devi, M. Singh, Kamlesh Mmit Hathras Computer Science and Engineering Uttar Pradesh India Prakasam Engineering College Science and Engineering Andhra Pradesh Kandukur India Srm Institute of Science and Technology Kattankulathur Campus School of Computing College of Engineering and Technology Department of Computing Technologies Chennai India Sri Krishna Adithya College of Arts and Science Department of Computer Applications Tamilnadu Coimbatore India St.Martin's Engineering College Telangana Secunderabad India Graphic Era Hill University Department of Computer Science and Engineering Dehradun Campus Dehradun India

ISBN: (纸本)9798350300093

The dissemination of news and coverage of events, both domestically and internationally, is now mostly the responsibility of print, web, and broadcast media. Because of the explosion of information available online, verifying the veracity of claims is more difficult than ever. It is extremely difficult to verify the veracity of fake news using conventional data processing techniques without seriously misinforming the public. Indeed, the difficulty for the government and the public is a matter of case by case argument. Several sites were built specifically for this purpose, each using its own unique logic and algorithm to determine which stories were more likely to be true or false. There needs to be a system in place for verifying the veracity of online claims, especially those that garner thousands of shares and clicks before being disproved by credible sources. In this proposed approach to an LSTM-RF classifier was used to determine whether or not a news story was fabricated. The three-step process that makes up the suggested method is as follows: preprocessing;feature extraction and feature selection;and LSTM-RF model training. In preprocessing, you'll clean the data, get rid of stop words, tokenize it, and stem it. The TF-IDF and Extra Tree Classifier are used for feature selection. The model is trained with LS TM-RF. The proposed method achieves superior results compared to LSTM and RF. © 2023 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Multi-modality Collaborative Recommender Systems: An Overview of Techniques and Evaluation Metrics

Multi-modality Collaborative Recommender Systems: An Overvie...

引用

2023 International Conference on Advanced Computing and Communication Technologies, ICACCTech 2023

作者： Gupta, Sorabh Bindal, Amit Kumar Prasad, Devendra M. M. Engineering College M. M. Deemed to Be University Department of Computer Science & Engineering Mullana India Panipat Institute of Engineering & Technology Department of Computer Science & Engineering Samalkha India

ISBN: (纸本)9798350380880

Multimodal Collaborative Recommendation Systems (MCRecSys) have emerged as essential solutions in the digital age, addressing the challenges posed by a vast array of online content and services. This paper offers an overview of the landscape, emphasizing the significance of multimodal content while outlining primary approaches. We categorize these models based on their techniques, ranging from traditional matrix factorization to advanced deep learning configurations. A particular emphasis is placed on multimodal data fusion, highlighting three methodologies: early, intermediate, and late fusion. Additionally, we delve into evaluation metrics vital for assessing system performance and discuss commonly used loss functions such as mean square error (MSE), Bayesian Personalized Ranking (BPR) loss, and contrastive loss. Moreover, this study elucidates the role of self-supervised learning and pretraining in enhancing recommendation results. This study outlines a coherent pipeline, assisting researchers in grasping the prevalent concepts and methodologies in the field. At its core, the progression of MCRecSys underscores the potential of leveraging diverse modalities to deliver more precise and personalized recommendations in today's digital ecosystem. © 2023 IEEE.

关键词： Recommender systems

来源：评论

学校读者我要写书评

暂无评论

Surface Defect Detection Algorithm for Strip Steel Based on Improved YOLOv7 Model

IAENG International Journal of Computer Science

引用

IAENG International Journal of computer science 2024年第3期51卷 308-316页

作者： Wang, Zhu Liu, Weisheng the College of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan China the College of Computer Science and Software Engineering the University of Science and Technology Liaoning CO Anshan114051 China

This research proposes a refined deep learning framework aimed at boosting the precision and efficacy of detecting surface imperfections in strip steel. This method integrates enhancement and simplification techniques inspired by the You Only Look Once version 7 (YOLOv7) detection method, resulting in significant enhancements in the model's accuracy, speed, and flexibility. The substitution of ELAN with Bottleneck Transformer 3 (BoT3) leads to improved accuracy and mean Average Precision (mAP) values, while also introducing a more lightweight network architecture. The incorporation of the Involution mechanism enhances the model's feature extraction capabilities, thereby improving its ability to recognize small targets through the utilization of local perceptual fields. The ASPP_CA architecture leverages a multi-scale feature fusion technique along with an attention mechanism to reduce model parameters and enhance inference speed. Furthermore, it extends the model's receptive field, allowing it to capture additional visual information. The enhanced algorithm, denoted as YOLOv7-IBA, demonstrates empirical results that underscore its superiority over the three current state-of-the-art detection techniques in identifying surface flaws on strip steel. The accuracy has been improved to 82.9%, representing a significant increase of 7.2% compared to the previous performance. Furthermore, the mean mAP value has experienced a 3.2% increase, reaching a total of 79.9%. Moreover, there has been a remarkable 8.8% improvement in efficiency. The adoption of this approach holds the potential to enhance both the precision and productivity of strip surface flaw detection, while also providing valuable methodological support for the advancement of other related disciplines. © (2024), (International Association of Engineers). All Rights Reserved.

关键词： Object detection

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：