检索结果-内蒙古大学图书馆

Viewport prediction, Bitrate Selection, and Beamforming Design for THz-Enabled 360° Video Streaming

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS 2025年第3期24卷 1849-1865页

作者： Setayesh, Mehdi Wong, Vincent W. S. Univ British Columbia Dept Elect & Comp Engn Vancouver BC V6T 1Z4 Canada

360 degrees videos require significant bandwidth to provide an immersive viewing experience. Wireless systems using terahertz (THz) frequency band can meet this high data rate demand. However, self-blockage is a challenge in such systems. To ensure reliable transmission, this paper explores THz-enabled 360 degrees video streaming through multiple multi-antenna access points (APs). Guaranteeing users' quality of experience (QoE) requires accurate viewport prediction to determine which video tiles to send, followed by asynchronous bitrate selection for those tiles and beamforming design at the APs. To address users' privacy and data heterogeneity, we propose a content-based viewport prediction framework, wherein users' head movement prediction models are trained using a personalized federated learning (PFL) algorithm. To address asynchronous decision-making for tile bitrates and dynamic THz link connections, we formulate the optimization of bitrate selection and beamforming as a macro-action decentralized partially observable Markov decision process (MacDec-POMDP) problem. To efficiently tackle this problem for multiple users, we develop two deep reinforcement learning (DRL) algorithms based on multi-agent actor-critic methods and propose a hierarchical learning framework to train the actor and critic networks. Experimental results show that our proposed approach provides a higher QoE when compared with three benchmark algorithms.

关键词： Streaming media prediction algorithms Bit rate Array signal processing Wireless communication Quality of experience Predictive models Terahertz communications Accuracy Solid modeling Deep reinforcement learning (DRL) macro-action decentralized partially observable Markov decision process (MacDec-POMDP) personalized federated learning (PFL) quality of experience (QoE) terahertz (THz) communication 360 degrees video viewport prediction

来源：评论

学校读者我要写书评

暂无评论

Knowledge Graph Neural Network With Spatial-Aware Capsule for Drug-Drug Interaction prediction

引用

IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS 2025年第3期29卷 1771-1781页

作者： Su, Xiaorui Zhao, Bowei Li, Guodong Zhang, Jun Hu, Pengwei You, Zhuhong Hu, Lun Chinese Acad Sci Xinjiang Tech Inst Phys & Chem Urumqi 830011 Peoples R China Univ Illinois Dept Comp Sci Chicago IL 60607 USA Northwestern Polytech Univ Sch Comp Sci Xian 710129 Peoples R China

Uncovering novel drug-drug interactions (DDIs) plays a pivotal role in advancing drug development and improving clinical treatment. The outstanding effectiveness of graph neural networks (GNNs) has garnered significant interest in the field of DDI prediction. Consequently, there has been a notable surge in the development of network-based computational approaches for predicting DDIs. However, current approaches face limitations in capturing the spatial relationships between neighboring nodes and their higher-level features during the aggregation of neighbor representations. To address this issue, this study introduces a novel model, KGCNN, designed to comprehensively tackle DDI prediction tasks by considering spatial relationships between molecules within the biomedical knowledge graph (BKG). KGCNN is built upon a message-passing GNN framework, consisting of propagation and aggregation. In the context of the BKG, KGCNN governs the propagation of information based on semantic relationships, which determine the flow and exchange of information between different molecules. In contrast to traditional linear aggregators, KGCNN introduces a spatial-aware capsule aggregator, which effectively captures the spatial relationships among neighboring molecules and their higher-level features within the graph structure. The ultimate goal is to leverage these learned drug representations to predict potential DDIs. To evaluate the effectiveness of KGCNN, it undergoes testing on two datasets. Extensive experimental results demonstrate its superiority in DDI predictions and quantified performance.

关键词： Drugs Knowledge graphs Semantics prediction algorithms Representation learning Task analysis Tail Drug-drug interaction prediction spatial- aware capsules non-linear aggregator biomedical knowledge graph graph neural network

来源：评论

学校读者我要写书评

暂无评论

Innovative Tailored Semantic Embedding and Machine Learning for Precise prediction of Drug-Drug Interaction Seriousness

引用

IEEE ACCESS 2025年 13卷 49249-49270页

作者： Mostafa, Ayman Mohamed Alaerjan, Alaa S. Allahem, Hisham Aldughayfiq, Bader Alruily, Meshrif Tantawy, Alshaimaa A. Ezz, Mohamed Jouf Univ Coll Comp & Informat Sci Informat Syst Dept Sakaka 72388 Saudi Arabia Jouf Univ Coll Comp & Informat Sci Dept Comp Sci Sakaka 72388 Saudi Arabia Zagazig Univ Fac Comp & Informat Dept Informat Syst Zagazig 44519 Egypt

This study explores applying advanced machine-learning strategies, particularly improved semantic vectors, to predict the severity of drug-drug interactions (DDIs), a crucial element in pharmacovigilance. Based on the Adverse Event Reporting System (FAERS), our study aims to analyze the combination of advanced embedding techniques with state-of-the-art machine learning (ML) algorithms to identify and quantify DDI severity. The CatBoost Classifier is the center of our analysis, as it has emerged as the most effective model in the examined trials. We improved the performance by increasing the BioWordVec Indication Substance embedding specificity, a new creation constructed through transfer learning methodologies employed on the BioWordVec model. This approach employs not only the names of the drugs but also the indications for the drugs and the active substances, forming a highly semantic network capable of capturing multiple relations between drugs. Applying BioWordVec Indication Substance embedding combined with the CatBoost Classifier, especially using the contact-vectors method, provided the best F1 score of 73. 32% and an ROC AUC score of 84%. The results imply that this method effectively models and predicts severe consequences of DDIs using deep learning that comprehensively covers pharmacological and clinical aspects. Based on our results, we suggest incorporating semantic embedding and ML into the pharmacovigilance processes to improve the predictive potential of DDI evaluations. Thus, by enhancing the body of knowledge related to the analytical methods of assessing drug interactions, the present study substantially enhances the quality of clinical decision-making and patient protection. The novel embedding marks a significant step forward in the methodology, providing a more solid tool for the fine-grained dissection of the complexities needed in modern medicine, where multiple drug therapies are now the norm.

关键词： Drugs Biological system modeling Semantics Predictive models Deep learning Databases Transfer learning Adaptation models prediction algorithms Graph neural networks drug-drug interaction (DDI) CatBoost classifier BioWordVec semantic embedding

来源：评论

学校读者我要写书评

暂无评论

Adaptive Broad Learning Network for Lightweight Multi-Dimensional Spectrum prediction

引用

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY 2025年第3期74卷 4498-4509页

作者： Ji, Niancong Zhang, Yibin Guo, Yongan Wang, Yu Guo, Lantu Lin, Yun Mehbodniya, Abolfazl Gui, Guan Sari, Hikmet Nanjing Univ Posts & Telecommun Coll Telecommun & Informat Engn Nanjing 210003 Peoples R China Nanjing Univ Posts & Telecommun Jiangsu Key Lab Intelligent Informat Proc & Commun Nanjing 210003 Peoples R China China Res Inst Radiowave Propagat Qingdao 266107 Peoples R China Beijing Inst Technol Sch Informat & Elect Beijing 100811 Peoples R China Harbin Engn Univ Coll Informat & Commun Engn Harbin 150000 Peoples R China Kuwait Coll Sci & Technol KCST Dept Elect & Commun Engn Doha Kuwait

Multi-dimensional spectrum prediction is pivotal for enabling spectrum sharing and dynamic spectrum access (DSA), which are critical for mitigating spectrum resource scarcity and enhancing wireless communication system performance. Traditional spectrum prediction techniques predominantly employ machine learning (ML) or deep learning (DL) approaches. However, ML-based methods often require manual feature extraction, leading to inefficiencies, while DL-based techniques typically demand computationally intensive graphics processing units (GPUs). To address these limitations, this paper proposes a lightweight multi-dimensional spectrum prediction model based on an adaptive broad learning network (ABLN). The model utilizes a sliding window approach for preprocessing historical spectrum data and establishes an input layer containing mapped feature nodes and enhancement nodes by randomly generating weights and biases. The weight parameters of the network are determined by solving the pseudo-inverse problem, and the network structure is extended by incremental learning method without necessitating retraining, effectively reducing the model's computational complexity. Furthermore, an adaptive node increment module is employed to quickly obtain the optimal hyperparameters. Experimental results on real-world spectrum datasets demonstrate that the proposed ABLN method consistently outperforms benchmark algorithms in terms of both prediction accuracy and computational efficiency across multi-step prediction tasks. Additionally, the model exhibits robust stability and adaptability in diverse scenarios.

关键词： Adaptation models Accuracy Incremental learning Computational modeling Dynamic spectrum access Predictive models Benchmark testing prediction algorithms Data models Internet of Things Dynamic spectrum access (DSA) multi-dimensional spectrum prediction adaptive broad learning network (ABLN) lightweight

来源：评论

学校读者我要写书评

暂无评论

Adaptive Multistep prediction With Sequence-to-Sequence (Seq2Seq) Models

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2025年 PP卷 PP页

作者： Kelley, Joseph Hagan, Martin Oklahoma State Univ Sch Elect & Comp Engn Stillwater OK 74078 USA

This brief demonstrates for the first time that the sequence-to-sequence (Seq2Seq) model is an adaptive multistep predictor. The Seq2Seq model is fixed-weight adaptive, which means that the model can adapt to time-varying behaviors without having to update its weights and biases. Instead, the learning algorithm is embedded into the recurrent neural network (RNN) decoder. This brief examines the Seq2Seq model's ability to adapt to time-varying behaviors using both simulated and experimental data, and it also identifies a mechanism within the model that enables the adaptation.

关键词： Decoding Adaptation models Predictive models Training Global Positioning System Adaptive systems Loading Data models Recurrent neural networks prediction algorithms Adaptive multistep prediction nonlinear autoregressive with exogenous (NARX) recurrent neural network (RNN) sequence to sequence (Seq2Seq)

来源：评论

学校读者我要写书评

暂无评论

Boosting GNN-Based Link prediction via PU-AUC Optimization

引用

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 2025年第4期37卷 1635-1649页

作者： Mao, Yuren Hao, Yu Cao, Xin Gao, Yunjun Yao, Chang Lin, Xuemin Zhejiang Univ Sch Software Technol Hangzhou 310027 Peoples R China Enmotech Co Ltd Sydney NSW 2113 Australia Univ New South Wales Sch Comp Sci & Engn Sydney NSW 2033 Australia Zhejiang Univ Coll Comp Sci & Technol Hangzhou 310027 Peoples R China Shanghai Jiao Tong Univ Antai Coll Econ & Manage ment Shanghai 200240 Peoples R China

Link prediction, which aims to predict the existence of a link between two nodes in a network, has various applications ranging from friend recommendation to protein interaction prediction. Recently, Graph Neural Network (GNN)-based link prediction has demonstrated its advantages and achieved the state-of-the-art performance. Typically, GNN-based link prediction can be formulated as a binary classification problem. However, in link prediction, we only have positive data (observed links) and unlabeled data (unobserved links), but no negative data. Therefore, Positive Unlabeled (PU) learning naturally fits the link prediction scenario. Unfortunately, the unknown class prior and data imbalance of networks impede the use of PU learning in link prediction. To deal with these issues, this paper proposes a novel model-agnostic PU learning algorithm for GNN-based link prediction by means of Positive-Unlabeled Area Under the Receiver Operating Characteristic Curve (PU-AUC) optimization. The proposed method is free of class prior estimation and able to handle the data imbalance. Moreover, we propose an accelerated method to reduce the operational complexity of PU-AUC optimization from quadratic to approximately linear. Extensive experiments back up our theoretical analysis and validate that the proposed method is capable of boosting the performance of the state-of-the-art GNN-based link prediction models.

关键词： Optimization prediction algorithms Estimation Training Predictive models Optimization methods Feature extraction Boosting Network topology Learning systems Link prediction graph neural network PU-AUC optimization

来源：评论

学校读者我要写书评

暂无评论

Blueberry Remaining Shelf-Life prediction Based on the PSO-CNN-BiLSTM-MHA Model

引用

IEEE ACCESS 2025年 13卷 43238-43248页

作者： Liu, Mengya Cheng, Xu Cao, Yu Zhou, Qian Liaoning Petrochem Univ Sch Informat Sci & Control Engn Fushun 113001 Peoples R China Shenyang Agr Univ Coll Econ & Management Shenyang 110866 Peoples R China Shenyang Agr Univ Food Sci Coll Shenyang 110866 Peoples R China

Remaining shelf-life prediction of blueberries is crucial in reducing economic losses and enhancing market competitiveness. To meet the demand for high accuracy in predicting the remaining shelf-life of blueberries, this paper proposes a fusion model (PSO-CNN-BiLSTM-MHA) that combines Particle Swarm Optimization (PSO), Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory network (BiLSTM), and Multi-Head Attention (MHA) mechanisms for predicting the remaining shelf-life of 'Emerald' blueberries under two temperature conditions, 4 degrees C and 25 degrees C. In this study, seven key features were selected from fifteen quality indicators of blueberries using the embedded method, and the PSO algorithm was used to determine the optimal hyperparameter combination of the model, which effectively improved its prediction performance. The experimental results show that our model outperforms other models in all evaluation metrics under both temperature conditions and demonstrates excellent prediction ability and stability. This study provides an effective technical approach for the accurate prediction of the remaining shelf-life of blueberries and other fruits.

关键词： Predictive models Data models Feature extraction Accuracy Kinetic theory Convolutional neural networks Time series analysis prediction algorithms Data mining Deep learning Bidirectional long short-term memory network convolutional neural network multi-head attention mechanisms particle swarm optimization remaining shelf-life prediction

来源：评论

学校读者我要写书评

暂无评论

HSoMLSDP: A Hybrid Swarm-Optimized Machine Learning Framework for Software Defect prediction

引用

IEEE ACCESS 2025年 13卷 61161-61180页

作者： Das, Madhusmita Mohan, Biju R. Guddeti, Ram Mohana Reddy Natl Inst Technol Karnataka Dept Informat Technol Surathkal 575025 Mangalore India

This research aims to design a hybrid swarm-optimized machine learning software defect prediction (HSoMLSDP) framework to predict software defects. We strive to do this by designing a swarm-optimized machine learning defect prediction (SoMLDP) model within the HSoMLSDP framework. In pursuit of enhancing the defect prediction accuracy of the SoMLDP model, this paper designed two novel hybrid swarm-optimization algorithms (SOAs) referred to as gravitational force grasshopper optimization algorithm-artificial bee colony (GFGOA-ABC), and levy flight grasshopper optimization algorithm-artificial bee colony (LFGOA-ABC) algorithms. By combining the enhanced exploration features of LFGOA and GFGOA with the robust exploitation capacity of the artificial bee colony (ABC), the LFGOA-ABC and GFGOA-ABC algorithms are proposed. Prior to validating the HSoMLSDP framework, the LFGOA-ABC and GFGOA-ABC algorithm's efficacy is first confirmed by experimenting on 19 benchmark functions (BFs) to assess their mean, standard deviation (SD) of optimal values, convergence rate, and convergence rate improvements. Following BFs verification, the second experiment tunes the hyperparameters of the ML models (artificial neural network, XGBOOST) to improve the defect accuracy of the SoMLDP model. The outcomes of the experiments justify a more rapid convergence rate for BFs and notable enhancements of 0.01-0.28 in software defect prediction (SDP) accuracy for NASA defect datasets when compared with state-of-the-art methods. As an enhancement of accuracy justifies the correctness of the SoMLDP model, thus validating the HSoMLSDP framework.

关键词： Optimization Tuning prediction algorithms Software Convergence Accuracy Software algorithms Predictive models Machine learning Machine learning algorithms Swarm-optimization algorithm benchmark functions machine learning software defect prediction

来源：评论

学校读者我要写书评

暂无评论

An Integrated prediction-Decision Unit Commitment Model Based on Deep Incremental Quantile Neural Networks

引用

IEEE ACCESS 2025年 13卷 39572-39584页

作者： Zhang, Han Wu, Tao Hu, Bo Shao, Changzheng Su, Yifan Ruan, Gang Chongqing Univ Sch Elect Engn Chongqing 400044 Peoples R China

Uncertainties and variabilities stemming from renewable energy make accurate prediction challenging, and thus pose significant difficulties for power system operation. This paper proposes an integrated prediction-decision unit commitment (IPD-UC) framework based on deep incremental quantile neural networks. First, the quantile distribution of the forecasting errors is obtained using deep incremental quantile neural networks. Error intervals subject to the pre-defined confidence requirements are selected using winkle coefficients. Then, a cost-oriented interval selection strategy is developed to obtain the economically optimal interval based on up-to-date information. Finally, an IPD-UC model is proposed to make optimal and robust decisions in response to uncertainties originating from renewable energy and the prediction algorithms. Case studies on the IEEE RTS 24-bus system and the IEEE 118-bus system demonstrate that the proposed integrated prediction-decision model reduces the total annual system cost by 4.2% and improves accuracy by 0.31%, showcasing its ability to enhance economic benefits and resilience.

关键词： Renewable energy sources Predictive models Optimization Neural networks Costs Load modeling Adaptation models Accuracy Uncertainty prediction algorithms Integrated prediction-decision unit commitment deep incremental quantile neural network renewable energy

来源：评论

学校读者我要写书评

暂无评论

Application Research of Multisource Data Fusion and Multimodel Ensemble Methods in Aircraft Approach State prediction

引用

IEEE SENSORS JOURNAL 2025年第5期25卷 8493-8505页

作者： Cao, Kang Zhang, Yongjie Feng, Jianfei Wang, Zhanchao Northwestern Polytech Univ Sch Civil Aviat Xian 710072 Shaanxi Peoples R China Shanghai Aircraft Design & Res Inst Shanghai 201210 Peoples R China Aviat Ind Dev Res Ctr China Beijing 100012 Peoples R China

With advances in sensing technology, situational awareness and prediction based on multisource data have become crucial for ensuring flight safety. Compared to single-model approaches, multimodel ensemble prediction algorithms offer higher accuracy and robustness, yet current ensemble strategies are insufficient for further development. Therefore, this article introduces a new multimodel ensemble prediction algorithm-the minimum consensus cost ensemble (MCCE) prediction algorithm-specifically for predicting the final approach state of civil aircraft. The MCCE algorithm incorporates the "minimum consensus cost" concept from group decision-making, quantifying the bias of individual models toward different prediction outcomes and dynamically adjusting the ensemble strategy to select the result with the highest consensus among models as the final output. This approach addresses decision bias issues in traditional ensemble models, enhancing stability and accuracy. In experiments, a fusion model was designed to maximize the silhouette coefficient, combining convolutional neural networks (CNNs) and multihead attention (MHA) mechanisms to efficiently integrate multisource data and isolate final approach state features. Through independent modeling and performance comparisons with multiple classification algorithms such as SVM and CNN, results show that the MCCE algorithm achieves approximately 99.97% accuracy, maintaining about 94.71% accuracy even under 20-dB signal-to-noise ratio (SNR) noise interference, significantly outperforming other models. Based on MCCE's consensus cost, the key parameters leading to abnormal approach states in civil aircraft can be identified, helping pilots make timely, effective flight decisions to reduce approach-phase accident risks.

关键词： Predictive models Atmospheric modeling prediction algorithms Data models Accuracy Sensors Convolutional neural networks Computer architecture Classification algorithms Accidents Aircraft approach classification group decision making multimodel ensemble multimodel fusion

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：