检索结果-内蒙古大学图书馆

Gasoline and Crude Oil Price Prediction using Multi-headed Variational Neighbour Search-tuned Recurrent Neural Networks

引用

Computational Economics 2025年 1-36页

作者： Kljajic, Maja Mizdrakovic, Vule Jovanovic, Luka Bacanin, Nebojsa Simic, Vladimir Pamucar, Dragan Zivkovic, Miodrag Business Economics Singidunum University Danijelova 32 Belgrade 11000 Serbia Department of Mathematics Saveetha School of Engineering SIMATS Thandalam Tamilnadu Chennai 602105 India Informatics and Computing Singidunum University Danijelova 32 Belgrade 11000 Serbia Sinergija University Raje Banjičića Bijeljina 76300 Bosnia and Herzegovina University of Belgrade Faculty of Transport and Traffic Engineering Vojvode Stepe 305 Belgrade 11010 Serbia Yuan Ze University College of Engineering Department of Industrial Engineering and Management Taoyuan City 320315 Taiwan Department of Computer Science and Engineering College of Informatics Korea University Seoul 02841 South Korea Department of Operations Research and Statistics Faculty of Organizational Sciences University of Belgrade Belgrade Serbia Department of Mechanics and Mathematics Western Caspian University Baku Azerbaijan İstinye University Faculty of Engineering and Natural Sciences Industrial Engineering Department İstanbul Turkey

The rapidly growing worldwide demand for energy and finite reservoirs of fossil fuels have intensified interest in energy projection research. Artificial intelligence, particularly in time series forecasting, holds significant promise for enhancing predictions of both cost and demand, offering numerous prospective applications over various domains. Numerous variables, ranging from socio-economic conditions to distribution, supply, and international policies, exert influence on global price fluctuations. Thus, crafting precise forecasts necessitates consideration of these multifaceted factors. Through an examination of existing literature, a discernible gap emerges in the quest for advancement in this domain. Consequently, this study proposes to delve into the perspective of employing multi-headed long short-term memory (LSTM) models for gasoline and crude oil price prediction, an area largely untouched by multi-headed approaches. Moreover, recognizing the computational demands of such models, this research emphasizes the development of lightweight methodologies, characterized by a modest neuron within each layer and trained over a limited epoch count. Given the pivotal role of hyper-parameter selection in algorithm performance, an adapted version of the variable neighbour search algorithm is introduced to aid in tuning the model’s architecture and training parameters. A comprehensive side-by-side comparison is undertaken utilizing gasoline and oil data sourced from diverse public repositories, employing a variety of contemporary optimizers. The resultant outcomes are subjected to strict statistical scrutiny to ascertain the robustness and significance of the findings. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.

关键词： Crude oil prices Gasoline prices Modified VNS Multi-headed long short-term memory models Time series prediction

来源：评论

学校读者我要写书评

暂无评论

Sustainable energy management in the AI era: a comprehensive analysis of ML and DL approaches

引用

Computing 2025年第6期107卷 1-64页

作者： Javed, Haseeb Eid, Fatma El-Sappagh, Shaker Abuhmed, Tamer Department of Computer Science and Engineering College of Computing and Informatics Sungkyunkwan University Suwon South Korea Technology Management Stony Brook University New York USA Applied Artificial Intelligence College of Computing and Informatics Sungkyunkwan University Suwon South Korea Faculty of Computer Science and Engineering Galala University Suez Egypt Information Systems Department Faculty of Computers and Artificial Intelligence Benha University Banha Egypt

This study comprehensively analyzes the application of innovative deep learning (DL) and machine learning (ML) techniques in smart energy management systems (EMSs), with an emphasis on load forecasting, demand response, and the development of smart energy sectors. The application of various ML and DL models were examined in over 200 studies from 2014 to 2024 in an electrical network's EMS to highlight the key benefits and advances made by each technology for the sustainable management systems in energy sector. The findings emphasize DL and ML models’ enhanced precision and predictive capabilities in load forecasting, their efficacy in enabling efficient demand response mechanisms, and their significance in supporting the development of smart energy sectors. Furthermore, recommendations are made based on the survey results to assist in incorporating these techniques into EMS frameworks, such as investment in data infrastructure, model training and validation, and collaboration between researchers, industry experts, and policymakers. The study also discusses the limitations identified in the literature, such as limited real-world implementations, challenges regarding quality and data availability, and the need for enhanced ML and DL model interpretability. Addressing these limitations can assist in increasing the application and efficacy of ML and DL techniques in EMSs, enabling a more efficient and sustainable energy landscape. Finally, this study facilitates researchers' exploration of ML and DL in energy management, highlighting relevant limitations, strengths, and alternative approaches associated with sustainable energy management. It also indicates potential future research directions for further investigation.

关键词： Power management (telecommunication)

来源：评论

学校读者我要写书评

暂无评论

YNote: A Novel Music Notation for Fine-Tuning LLMs in Music Generation

arXiv

引用

arXiv 2025年

作者： Lu, Shao-Chien Yeh, Chen-Chen Cho, Hui-Lin Hsu, Chun-Chieh Hsu, Tsai-Ling Wu, Cheng-Han Shih, Timothy K. Lin, Yu-Cheng Department of Computer Science and Engineering Yuan Ze University Taoyuan Taiwan International Bachelor Program in Informatics Yuan Ze University Taoyuan Taiwan Department of Computer Science and Information Engineering National Central University Taoyuan Taiwan

The field of music generation using Large Language Models (LLMs) is evolving rapidly, yet existing music notation systems, such as MIDI, ABC Notation, and MusicXML, remain too complex for effective fine-tuning of LLMs. These formats are difficult for both machines and humans to interpret due to their variability and intricate structure. To address these challenges, we introduce YNote, a simplified music notation system that uses only four characters to represent a note and its pitch. YNote’s fixed format ensures consistency, making it easy to read and more suitable for fine-tuning LLMs. In our experiments, we fine-tuned GPT-2 (124M) on a YNote-encoded dataset and achieved BLEU and ROUGE scores of 0.883 and 0.766, respectively. With just two notes as prompts, the model was able to generate coherent and stylistically relevant music. We believe YNote offers a practical alternative to existing music notations for machine learning applications and has the potential to significantly enhance the quality of music generation using LLMs. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Automatic Measurement of Hoof Conformation in Horses

Automatic Measurement of Hoof Conformation in Horses

引用

Medical Imaging 2025: Imaging informatics

作者： Mackwan, Brian S. Haas, Paulo Deserno, Verena Deserno, Thomas M. Indian Institute of Technology Bombay Department of Computer Science & Engineering Mumbai India Peter L. Reichertz Institute for Medical Informatics of TU Braunschweig Hannover Medical School Braunschweig Germany Huforthopädie und Pferdephysiotherapie Peine Peine Germany

ISBN: (纸本)9781510686007

Veterinarians can use photographs of horse hooves to detect abnormalities and assess their morphology and conformation. However, manual segmentation of these images is time-consuming. Automatization reduces the workload and makes the conformation assessment accessible to everybody. In this work, we use 280 images of seven horses labeled with four key points. We train a ResNet50 model with cross-validation, leaving one horse out. The model automatically predicts four key points. We reject 2.56 % of the tested images as prediction failures due to the distance of predicted and ground truth key points. With the rest, we calculate dorsal hoof length, support length, heel height, toe height, hoof angle, coronary band angle, and heel angle. On average, the deviations are −0.75 %, 0.5 %, −0.24 %, −0.21 %, −0.19◦, −0.07◦, 0.04◦ between the prediction and the ground truth, respectively. The agreement indices between the prediction and ground truth are 95.62 %, 96.97 %, 99.02 %, 99.30 %, 96.03 %, 88.98 %, and 95.68 %, respectively. The intersection over unit is 91.66 %. Neural networks can detect the key points of horse hooves accurately. Our model yields low deviations for lengths and angles on average, but faces limitations in determining angles reliably. © 2025 SPIE

关键词： Veterinary medicine

来源：评论

学校读者我要写书评

暂无评论

Joint Content Caching and Request Routing for User-Centric Many-Objective Metaverse Services

引用

IEEE Transactions on Network Science and engineering 2025年第3期12卷 1911-1925页

作者： Hu, Zhaoming Fang, Chao Wang, Zhuwei Chen, Jining Tseng, Shu-Ming Dong, Mianxiong Beijing University of Technology College of Computer Science Beijing100021 China Beijing University of Technology School of Information Science and Technology Beijing100021 China Guangxi Information Center Guangxi Key Laboratory of Digital Infrastructure Nanning530000 China National Taipei University of Technology Department of Electronic Engineering Taipei106 Taiwan Muroran Institute of Technology Department of Sciences and Informatics Muroran050-0071 Japan

Metaverse, as a revolutionary technology that changes the way of human interaction, brings new challenges to content delivery services due to the extensive data transmission and personalized service requirements. To ensure a personalized user experiences while improving the utilization of heterogeneous network resources, a user-centric many-objective metaverse content delivery framework is proposed to optimize content delivery through user attention awareness. This framework addresses two key subproblems in metaverse content delivery by investigating user-centric many-objective cooperative content caching and deep reinforcement learning (DRL)-based request routing. The user-centric many-objective cooperative content caching is proposed to dynamically combine three basic preference prediction results to predict user preferences and control network resource allocation, which can simultaneously optimize prediction precision, delay, offloaded traffic, and load balancing. In DRL-based request routing, the reward function is designed to enable the optimization of multiple objectives. The multi-objective DRL routing algorithm is employed to continuously observe network states and make adaptive routing decisions in response to user requests. In the simulation, a movie dataset is employed to simulate user requests and support user attention awareness. The results show that the proposed content delivery framework outperforms existing basic prediction algorithms and other content delivery algorithms on four evaluation indicators. © 2013 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Leisure time under pressure: Exploring the impact of professional and personal demands on female Indonesian lecturers

引用

Journal of Hospitality, Leisure, Sport and Tourism Education 2025年 36卷

作者： Astuty, Sri Marhawati Hastuti, Diah Retno Dwi Setialaksana, Wirawan Economics Department Universitas Negeri Makassar Indonesia Informatics and Computer Engineering Education Universitas Negeri Makassar Indonesia

This study examines the impact of age, education, performance, caregiving responsibilities, and leisure time among 302 female Indonesian lecturers using PLS-SEM. Results show that age and education increase productivity and income but reduce leisure time, while higher performance lowers both income and leisure time. The presence of a toddler mitigates performance's negative effect on leisure, while education buffers its impact on income. Findings highlight work-life balance challenges, emphasizing the need for flexible workloads and institutional support. Despite its cross-sectional design, this study offers insights into gendered work structures in academia, particularly in emerging markets. © 2025 Elsevier Ltd

关键词： Dependents Mediation analysis Moderation analysis Presence of a toddler Structural equation model

来源：评论

学校读者我要写书评

暂无评论

Accelerated Image Stitching Via Parallel Computing for UAV Applications

Accelerated Image Stitching Via Parallel Computing for UAV A...

引用

AIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2025

作者： Ramirez, Rick Korah, John Bhandari, Subodh Nguyen, Tu Chen, Yuqi Le, Du D. Computer Science Department California State Polytechnic University PomonaCA91768 United States Aerospace Engineering Department California State Polytechnic University PomonaCA91768 United States Computer Science Department University of California IrvineCA92697 United States Informatics Department University of California IrvineCA92697 United States Mathematics and Computer Science Department Mt. San Antonio College CA91789 United States

ISBN: (数字)9781624107238

ISBN: (纸本)9781624107238

In this work, we formulate an accelerated image fusion algorithm for Unmanned Aerial Vehicle (UAV) application which is based on image stitching using invariant features. By utilizing parallel computing techniques for distributed and shared memory architectures, we have improved both the run time of the feature-based image stitching procedure, and the quality of the results when compared to sequential processing. The images used to generate the panorama were obtained from a UAV at a frame rate of 30Hz and a resolution of 3840×2160. As the UAV traverses across the landscape, a video frame buffer is collected and sent to a multi-core Central Processing Unit (CPU) where individual CPU cores perform image fusion in parallel. Our algorithm can be applied to any multi-core CPU (or cluster of multi-core CPU architectures) with support for Graphics Processing Unit (GPU) accelerations. The procedure was tested on various hardware platforms including edge computing devices, ground stations, and high-performance computing environments. The results show that as we increase the number of available CPU cores, there is a decrease in the runtime by as much as 7x under the distributed memory design. There is also a decrease in runtime during the feature detection stage of the fusion process under the shared memory design by a factor of 10x. Finally, by incorporating GPU acceleration at various stages of the stitching procedure, we can decrease the runtime by another factor of 2x. Furthermore, as we increase the number of threads utilized in the shared memory design, we find that there is a significant improvement in the quality of the panorama. © 2025, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

FEDERATED INVERSE PROBABILITY TREATMENT WEIGHTING FOR INDIVIDUAL TREATMENT EFFECT ESTIMATION

arXiv

引用

arXiv 2025年

作者： Yin, Changchang Chen, Hong-You Chao, Wei-Lun Zhang, Ping Department of Computer Science and Engineering The Ohio State University ColumbusOH United States Department of Computer Science and Engineering Department of Biomedical Informatics Translational Data Analytics Institute The Ohio State University ColumbusOH United States

Individual treatment effect (ITE) estimation is to evaluate the causal effects of treatment strategies on some important outcomes, which is a crucial problem in healthcare. Most existing ITE estimation methods are designed for centralized settings. However, in real-world clinical scenarios, the raw data are usually not shareable among hospitals due to the potential privacy and security risks, which makes the methods not applicable. In this work, we study the ITE estimation task in a federated setting, which allows us to harness the decentralized data from multiple hospitals. Due to the unavoidable confounding bias in the collected data, a model directly learned from it would be inaccurate. One well-known solution is Inverse Probability Treatment Weighting (IPTW), which uses the conditional probability of treatment given the covariates to re-weight each training example. Applying IPTW in a federated setting, however, is non-trivial. We found that even with a well-estimated conditional probability, the local model training step using each hospital’s data alone would still suffer from confounding bias. To address this, we propose FED-IPTW, a novel algorithm to extend IPTW into a federated setting that enforces both global (over all the data) and local (within each hospital) decorrelation between covariates and treatments. We validated our approach on the task of comparing the treatment effects of mechanical ventilation on improving survival probability for patients with breadth difficulties in the intensive care unit (ICU). We conducted experiments on both synthetic and real-world eICU datasets and the results show that FED-IPTW outperform state-of-the-art methods on all the metrics on factual prediction and ITE estimation tasks, paving the way for personalized treatment strategy design in mechanical ventilation usage. © 2025, CC BY-NC-SA.

关键词： Inverse problems

来源：评论

学校读者我要写书评

暂无评论

Recursive Broadcasting Approach

Recursive Broadcasting Approach

引用

Euromicro Conference on Parallel, Distributed and Network-Based Processing

作者： Hovhannes A. Harutyunyan Narek Hovhannisyan Department of Computer Science and Software Engineering Concordia University Montreal QC Canada NAS RA Institute for Informatics and Automation Problems Yerevan Armenia

ISBN: (数字)9798331524937

ISBN: (纸本)9798331524944

Broadcasting is one of the fundamental information dissemination primitives in interconnection networks, where a message is passed from one node (called originator) to all other nodes in the network. Following the increasing interest in interconnection networks, extensive research was dedicated to broadcasting. Two main research goals of this area are finding inexpensive network structures that maintain efficient broadcasting and finding the broadcast time for well-known and widely used network topologies. In the scope of this study, we will mainly focus on determining the broadcast time and nearoptimal broadcasting schemes in networks. Determination of the broadcast time of any node in an arbitrary network is known to be NP-hard. Polynomial time solutions are known only for a few network topologies. There also exist various heuristic and approximation algorithms for different network topologies. In this study, we consider the broadcast time problem on graphs that comprise some recursive structures. We initiate a novel direction to designing broadcasting algorithms on recursively defined graphs. We provide a theoretical foundation for future broadcasting studies, as well as discuss several practical applications of the approach we introduce.

关键词： Multicast algorithms Network topology Multiprocessor interconnection Heuristic algorithms Broadcasting Approximation algorithms Polynomials Topology

来源：评论

学校读者我要写书评

暂无评论

Late Breaking Results: Weight Decay is All You Need for Neural Network Sparsification 23

Late Breaking Results: Weight Decay is All You Need for Neur...

引用

Proceedings of the 60th Annual ACM/IEEE Design Automation Conference

作者： Xizi Chen Rui Pan Xiaomeng Wang Fengshi Tian Chi-Ying Tsui College of Informatics Huazhong Agricultural University China Department of Electronic and Computer Engineering The Hong Kong University of Science and Technology Hong Kong China

ISBN: (纸本)9798350323481

The heuristic iterative pruning strategy has been widely used for neural network sparsification. However, it is challenging to identify the right connections to remove at each pruning iteration with only a one-shot evaluation of weight magnitude, especially at the early pruning stage. The erroneously removed connections, unfortunately, can hardly be recovered. In this work, we propose a weight decay strategy as a substitute for pruning, which let the "insignificant" weights moderately decay instead of being directly clamped to zero. At the end of the training, the vast majority of redundant weights will naturally become close to zero, making it easier to identify which connections could be removed safely. Experimental results show that the proposed weight decay method can achieve an ultra-high sparsity of 99%. Compared to the current pruning strategy, the model size is further reduced by 34%, improving the compression rate from 69× to 106× at the same accuracy.

关键词： DNN

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：