检索结果-内蒙古大学图书馆

Multi-task learning andjoint refinement between camera localization and object detection

Computational Visual Media 2024年第5期10卷 993-1011页

作者： Junyi Wang Yue Qi State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and EngineeringBeihang UniversityBeijing 100191China Peng Cheng Laboratory Shenzhen 518052China Qingdao Research Institute of Beihang University Qingdao 266104China School of Computer Science and Technology Shandong UniversityQingdaoChina

Visual localization and object detection both play important roles in various *** many indoor application scenarios where some detected objects have fixed positions,the two techniques work closely ***,few researchers consider these two tasks simultaneously,because of a lack of datasets and the little attention paid to such *** this paper,we explore multi-task network design and joint refinement of detection and *** address the dataset problem,we construct a medium indoor scene of an aviation exhibition hall through a semi-automatic *** dataset provides localization and detection information,and is publicly available at https://***/drive/folders/1U28zk0N4_I0db zkqyIAK1A15k9oUKOjI?usp=sharing for benchmarking localization and object detection *** this dataset,we have designed a multi-task network,JLDNet,based on YOLO v3,that outputs a target point cloud and object bounding *** dynamic environments,the detection branch also promotes the perception of *** includes image feature learning,point feature learning,feature fusion,detection construction,and point cloud ***,object-level bundle adjustment is used to further improve localization and detection *** test JLDNet and compare it to other methods,we have conducted experiments on 7 static scenes,our constructed dataset,and the dynamic TUM RGB-D and Bonn *** results show state-of-the-art accuracy for both tasks,and the benefit of jointly working on both tasks is demonstrated.

关键词： visual localization object detection joint optimization multi-task learning

来源：评论

学校读者我要写书评

暂无评论

Towards Green AI by Reducing Training Effort of Recurrent Neural Networks Using Hyper-Parameter Optimization with Dynamic Stopping Criteria 22

Towards Green AI by Reducing Training Effort of Recurrent Ne...

引用

22nd IEEE International Symposium on Intelligent systems and Informatics, SISY 2024

作者： Podgorelec, Vili Fister, Iztok Vrbancic, Grega University of Maribor Intelligent Systems Laboratory Faculty of Electrical Engineering and Computer Science Slovenia

ISBN: (纸本)9798350385601

Neural networks have become a leading model in modern machine learning, able to model even the most complex data. For them to be properly trained, however, a lot of computational resources are required. With the carbon footprint of ever-growing adoption of neural networks in mind, an approach to reduce the required training resources would be very welcome. We designed a new training effort reduction method based on the calculation of area under the normalized loss curve and assessed it on the electricity consumption forecasting problem with the recurrent neural networks. The results show that the proposed method was able to considerably reduce the amount of computational resources, while maintaining the predictive performance, and thus contributing towards the Green AI. © 2024 IEEE.

关键词： Green AI hyper-parameter optimization machine learning neural networks resource optimization

来源：评论

学校读者我要写书评

暂无评论

Computationally efficient reductions between some statistical models 36

Computationally efficient reductions between some statistica...

引用

36th International Conference on Algorithmic Learning Theory, ALT 2025

作者： Lou, Mengqi Bresler, Guy Pananjady, Ashwin School of Industrial and Systems Engineering Georgia Tech United States Department of Electrical Engineering and Computer Science MIT United States Schools of Industrial and Systems Engineering and Electrical and Computer Engineering Georgia Tech United States

We study the problem of approximately transforming a sample from a source statistical model to a sample from a target statistical model without knowing the parameters of the source model, and construct several computationally efficient such reductions between canonical statistical experiments. In particular, we provide computationally efficient procedures that approximately reduce uniform, Erlang, and Laplace location models to general target families. We illustrate our methodology by establishing nonasymptotic reductions between some canonical high-dimensional problems, spanning mixtures of experts, phase retrieval, and signal denoising. Notably, the reductions are structure-preserving and can accommodate missing data. We also point to a possible application in transforming one differentially private mechanism to another. © 2025 M. Lou, G. Bresler & A. Pananjady.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

Supervised learning techniques for blood product prediction in patients with hematologic diseases: a multi-centre study in Western Algeria

引用

International Journal of Information Technology (Singapore) 2024年 1-12页

作者： Ykhlef, Amel Labri, Nedjla Selma Brahami, Menaouer LABAB Laboratory Computer Systems Engineering Department National Polytechnic School of Oran Oran Algeria

Blood transfusion is a medical procedure that involves transfusing blood or one of its components from one or more donors into a patient. Digital technology and machine learning have played a crucial role in the blood field and have provided real prospects for the production and distribution of blood products. In this study, we propose supervised machine learning techniques for the multi-label classification of blood products in patients with hematologic diseases. We used three multi-label approaches from the problem transformation category to create a decision support system for blood products label power set (LP), binary relevance (BR), and classifier chain (CC). Multi-label classification using the problem transformation approach is a flexible approach. In this study, we used data from different hospitals in hematology departments and blood transfusion centers to explore the application of contemporary supervised learning algorithms to blood product prediction. The experiment was performed by calculating the Hamming loss and accuracy to facilitate the classification of blood products. As a result, the prediction model achieved an area under the ROC curve of 99.80%, a Hamming loss of 0.30, and an accuracy of 98.8%. The proposed model has been developed to provide accurate and fast results that can save patients’ lives. © Bharati Vidyapeeth's Institute of computer Applications and Management 2024.

关键词： Blood product Blood transfusion Hematological diseases Knowledge management Machine learning Multi-label classification Problem transformation methods

来源：评论

学校读者我要写书评

暂无评论

MIMO performance enhancement of MIMO arrays using PCS-based near-field optimization technique

引用

Science China(Information Sciences) 2023年第6期66卷 170-185页

作者： Yipeng WANG Xiaoming CHEN Huiling PEI Wei E.I.SHA Yi HUANG Ahmed A.KISHK School of Information and Communications Engineering Xi'an Jiaotong University Key Laboratory of Micro-Nano Electronic Devices and Smart Systems of Zhejiang Province College of Information Science and Electronic EngineeringZhejiang University Department of Electrical Engineering and Electronics University of Liverpool Department of Electrical and Computer Engineering Concordia University

Extensive efforts have been made in designing large multiple-input multiple-output(MIMO)arrays. Nevertheless, improvements in conventional antenna characteristics cannot ensure significant MIMO performance improvement in realistic multipath environments. Array decorrelation techniques have been proposed, achieving correlation reductions by either tilting the antenna beams or shifting the phase centers away from each other. Hence, these methods are mainly limited to MIMO terminals with small arrays. To avoid such problems, this work proposes a decorrelation optimization technique based on phase correcting surface(PCS)that can be applied to large MIMO arrays, enhancing their MIMO performances in a realistic(non-isotropic)multipath environment. First, by using a near-field channel model and an optimization algorithm, a near-field phase distribution improving the MIMO capacity is obtained. Then the PCS(consisting of square elements)is used to cover the array's aperture, achieving the desired near-field phase *** examples demonstrate the effectiveness of this PCS-based near-field optimization technique. One is a1 × 4 dual-polarized patch array(working at 2.4 GHz)covered by a 2 × 4 PCS with 0.6λ center-to-center distance. The other is a 2 × 8 dual-polarized dipole array, for which a 4 × 8 PCS with 0.4λ center-to-center distance is designed. Their MIMO capacities can be effectively enhanced by 8% and 10% in single-cell and multi-cell scenarios, respectively. The PCS has insignificant effects on mutual coupling, matching, and the average radiation efficiency of the patch array, and increases the antenna gain by about 2.5 dB while keeping broadside radiations to ensure good cellular coverage, which benefits the MIMO performance of the *** proposed technique offers a new perspective for improving large MIMO arrays in realistic multipath in a statistical sense.

关键词： antenna correlation array antenna multiple-input multiple-output (MIMO) capacity

来源：评论

学校读者我要写书评

暂无评论

UKF‐MOT:An unscented Kalman filter‐based 3D multi‐object tracker

引用

CAAI Transactions on Intelligence Technology 2024年第4期9卷 1031-1041页

作者： Meng Liu Jianwei Niu Yu Liu Collective Intelligence&Collaboration Laboratory China North Artificial Intelligence and Innovation Research InstituteBeijingChina China North Vehicle Research Institute BeijingChina State Key Laboratory of Software Development Environment School of Computer Science and EngineeringBeihang UniversityBeijingChina State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and EngineeringBeihang UniversityBeijingChina School of Computer Science and Engineering Beihang UniversityBeijingChina

Multi‐object tracking in autonomous driving is a non‐linear *** better address the tracking problem,this paper leveraged an unscented Kalman filter to predict the object's *** the association stage,the Mahalanobis distance was employed as an affinity metric,and a Non‐minimum Suppression method was designed for *** the detections fed into the tracker and continuous‘predicting‐matching’steps,the states of each object at different time steps were described as their own continuous *** conducted extensive experiments to evaluate tracking accuracy on three challenging datasets(KITTI,nuScenes and Waymo).The experimental results demon-strated that our method effectively achieved multi‐object tracking with satisfactory ac-curacy and real‐time efficiency.

关键词： autonomous vehicle transportation

来源：评论

学校读者我要写书评

暂无评论

A Data-Driven Adaptive Control Approach for Enhancing the Dynamic Response of VSGs in Varying Grid Conditions

引用

IEEE Transactions on Power Delivery 2025年第3期40卷 1421-1433页

作者： Fahad, Shah She, Buxin Yin, Junjie Li, Fangxing Cui, Hantao Bo, Rui Department of Electrical and Computer Engineering Missouri University of Science and Technology Rolla MO USA Senior Power Systems Engineer ETAP Irvine CA USA Optimization & Control Group Parcific Northwest National Laboratory Richland USA Department of Electrical Engineering and Computer Science University of Tennessee at Knoxville Knoxville TN USA Department of Electrical and Computer Engineering North Carolina State University Raleigh NC USA

Conventionally, a virtual synchronous generator (VSG) is designed for islanded mode (IM) operation to meet specific operational requirements such as the rate of change of frequency (RoCoF). However, the operation of VSG designed for IM may not meet the operational and control criteria in grid connected mode (GCM) when the grid conditions vary. In addition, conventional VSG control technology does not consider the influence of the presynchronization scheme when connected to a weak grid, which degrades the RoCoF in IM. To overcome the aforementioned challenges, the proposed study presents a twin-delayed deep deterministic policy gradient (TD3) algorithm to improve the seamless transition performance of VSG from IM to GCM and vice versa. In the first step, the VSG-based power system model is used as a foundation for problem formulation of the proposed TD3 algorithm. Secondly, a reward function is designed according to the performance requirements, i.e., frequency and RoCoF requirements, of the VSG in order to guide the training of the agent in varying load, power reference, and grid conditions. Finally, the superiority of the proposed algorithm over existing methods is validated in MATLAB/SIMULINK and RTDS based real-time simulation environment. © 1986-2012 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

RISCORE: Enhancing In-Context Riddle Solving in Language Models through Context-Reconstructed Example Augmentation 31

RISCORE: Enhancing In-Context Riddle Solving in Language Mod...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Panagiotopoulos, Ioannis Filandrianos, Giorgos Lymperaiou, Maria Stamou, Giorgos School of Electrical and Computer Engineering AILS Laboratory National Technical University of Athens Greece

ISBN: (纸本)9798891761964

Riddle-solving requires advanced reasoning skills, pushing Large Language Models (LLMs) to engage in abstract thinking and creative problem-solving, often revealing limitations in their cognitive abilities. In this paper, we examine the riddle-solving capabilities of LLMs using a multiple-choice format, exploring how different prompting techniques impact performance on riddles that demand diverse reasoning skills. To enhance results, we introduce RISCORE (RIddle Solving with COntext REcontruciton), a novel fully automated prompting method that generates and utilizes contextually reconstructed sentence-based puzzles in conjunction with the original examples to create few-shot exemplars. Our experiments demonstrate that RISCORE significantly improves the performance of language models in both vertical and lateral thinking tasks, surpassing traditional exemplar selection strategies across a variety of few-shot settings. © 2025 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Latent Variable and Classification Performance Analysis of Bird-Drone Spectrograms with Elementary Autoencoder

IEEE Transactions on Radar Systems

引用

IEEE Transactions on Radar systems 2024年 3卷 115-123页

作者： White, Daniel Jahangir, Mohammed Mishra, Amit Kumar Baker, Chris J. Antoniou, Michail University Of Birmingham School Of Electrical Electronic And Systems Engineering United Kingdom Aberystwyth University Department Of Computer Science United Kingdom

Deep learning with convolutional neural networks has been widely utilised in radar research concerning automatic target recognition. Maximising numerical metrics to gauge the performance of such algorithms does not necessarily correspond to model robustness against untested targets, nor does it lead to improved model interpretability. Approaches designed to explain the mechanisms behind the operation of a classifier on radar data are proliferating, but bring with them a significant computational and analysis overhead. This work uses an elementary unsupervised convolutional autoencoder (CAE) to learn a compressed representation of a challenging dataset of urban bird and drone targets and subsequently if apparent quality of the representation via preservation of class labels leads to better classification performance after a separate supervised training stage. It is shown that a CAE that reduces the features output after each layer of the encoder gives rise to the best drone vs bird classifier. A clear connection between unsupervised evaluation via label preservation in the latent space and subsequent classification accuracy after supervised fine-tuning is shown, supporting further efforts to optimise radar data latent representations to enable optimal performance and model interpretability. © 2023 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Granular classifier:Building traffic granules for encrypted traffic classification based on granular computing

引用

Digital Communications and Networks 2024年第5期10卷 1428-1438页

作者： Xuyang Jing Jingjing Zhao Zheng Yan Witold Pedrycz Xian Li State Key Laboratory on Integrated Services Networks School of Cyber EngineeringXidian UniversityChina Department of Electrical and Computer Engineering University of AlbertaCanada Institute for Future(IFF) Qingdao UniversityQingdaoChina

Accurate classification of encrypted traffic plays an important role in network ***,current methods confronts several problems:inability to characterize traffic that exhibits great dispersion,inability to classify traffic with multi-level features,and degradation due to limited training traffic *** address these problems,this paper proposes a traffic granularity-based cryptographic traffic classification method,called Granular Classifier(GC).In this paper,a novel Cardinality-based Constrained Fuzzy C-Means(CCFCM)clustering algorithm is proposed to address the problem caused by limited training traffic,considering the ratio of cardinality that must be linked between flows to achieve good traffic ***,an original representation format of traffic is presented based on granular computing,named Traffic Granules(TG),to accurately describe traffic structure by catching the dispersion of different traffic *** granule is a compact set of similar data with a refined boundary by excluding *** on TG,GC is constructed to perform traffic classification based on multi-level *** performance of the GC is evaluated based on real-world encrypted network traffic *** results show that the GC achieves outstanding performance for encrypted traffic classification with limited size of training traffic and keeps accurate classification in dynamic network conditions.

关键词： Encrypted traffic classification Semi-supervised clustering Granular computing Anomaly 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：