检索结果-内蒙古大学图书馆

SSRN 2024年

作者： Zheng, Wenhao Xiong, Bangshu Yi, Hui Au, Qiaofeng Chen, Jiujiu Key Laboratory of Jiangxi Province for Image Processing and Pattern Recognition Nanchang Hangkong University Nanchang330063 China China Helicopter Research and Development Institute China

Detecting blade tip point light sources based on airborne computer vision is a critical step in measuring blade tip distance for coaxial unmanned helicopters. However, detecting blade tip point light sources quickly and highly accurately in a helicopter airborne environment is a challenge due to the extremely tiny imaging size of the blade tip point light sources. To this end, we propose a FRE-YOLO for detecting blade tip point light sources based on the YOLO network structure. In this work, we focus on designing a Feature Refinement Extraction network (FRE), which can greatly enrich the feature information of small targets during Backbone downsampling. Secondly, we also design a new paradigm Smooth Gaussian Distance (SmoothGD) which measures the degree of target bounding box matching. SmoothGD is mainly applied to regression loss calculation and label assigner, which can effectively improve the small target detection accuracy. Meanwhile, we construct the Blade Tip Point Light (BTPL) dataset for the network training. In addition, in order to evaluate the effectiveness of the proposed method more fairly, this paper conducts the same experiments on BTPL and the small target dataset VisDrone on MMYOLO. A large amount of experimental data shows that the proposed method has a positive impact on the detection of tiny point sources at the tip of the blade. It is also at the advanced level in the field of generalized small object detection. © 2024, The Authors. All rights reserved.

关键词： Extraction

来源：评论

学校读者我要写书评

暂无评论

DeepRAG: Thinking to Retrieval Step by Step for Large Language Models

arXiv

引用

arXiv 2025年

作者： Guan, Xinyan Zeng, Jiali Meng, Fandong Xin, Chunlei Lu, Yaojie Lin, Hongyu Han, Xianpei Sun, Le Zhou, Jie Chinese Information Processing Laboratory Institute of Software Chinese Academy of Sciences China University of Chinese Academy of Sciences China Pattern Recognition Center WeChat AI Tencent Inc China

Large Language Models (LLMs) have shown remarkable potential in reasoning while they still suffer from severe factual hallucinations due to timeliness, accuracy, and coverage of parametric knowledge. Meanwhile, integrating reasoning with retrieval-augmented generation (RAG) remains challenging due to ineffective task decomposition and redundant retrieval, which can introduce noise and degrade response quality. In this paper, we propose DeepRAG, a framework that models retrieval-augmented reasoning as a Markov Decision Process (MDP), enabling strategic and adaptive retrieval. By iteratively decomposing queries, DeepRAG dynamically determines whether to retrieve external knowledge or rely on parametric reasoning at each step. Experiments show that DeepRAG improves retrieval efficiency while improving answer accuracy by 21.99%, demonstrating its effectiveness in optimizing retrieval-augmented reasoning. © 2025, CC BY.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A New Deep Fuzzy Based MSER Model for Multiple Document Images Classification 3rd

A New Deep Fuzzy Based MSER Model for Multiple Document Imag...

引用

3rd International Conference on pattern recognition and Artificial Intelligence, ICPRAI 2022

作者： Biswas, Kunal Shivakumara, Palaiahnakote Sivanthi, Sittravell Pal, Umapada Lu, Yue Liu, Cheng-Lin Ayub, Mohamad Nizam Bin Computer Vision and Pattern Recognition Unit Indian Statistical Institute Kolkata India Faculty of Computer Science and Information Technology University of Malaya Kuala Lumpur Malaysia Shanghai Key Laboratory of Multidimensional Information Processing East China Normal University Shanghai China National Laboratory of Pattern Recognition Institute of Automation of Chinese Academy of Sciences Beijing China School of Artificial Intelligence University of Chinese Academy Beijing China

ISBN: (纸本)9783031090363

Understanding document images uploaded on social media is challenging because of multiple types like handwritten, printed and scene text images. This study presents a new model called Deep Fuzzy based MSER for classification of multiple document images (like handwritten, printed and scene text). The proposed model detects candidate components that represent dominant information irrespective of the type of document images by combining fuzzy and MSER in a novel way. For every candidate component, the proposed model extracts distance-based features which result in proximity matrix (feature matrix). Further, the deep learning model is proposed for classification by feeding input images and feature matrix as input. To evaluate the proposed model, we create our own dataset and to show effectiveness, the proposed model is tested on standard datasets. The results show that the proposed approach outperforms the existing methods in terms of average classification rate. © 2022, Springer Nature Switzerland AG.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

An Integrated Clustering-based Approach for Aircraft Trajectory Generation Model 41

An Integrated Clustering-based Approach for Aircraft Traject...

引用

41st IEEE/AIAA Digital Avionics Systems Conference, DASC 2022

作者： Fang, Yuhang Li, Wei Fang, Quan Yang, Yang Cai, Kaiquan Beihang University School of Electronic and Information Engineering Beijing China Institute of Automation National Lab of Pattern Recognition Beijing China Beihang University Research Institute for Frontier Science Beijing China

ISBN: (数字)9781665486071

ISBN: (纸本)9781665486071

Accurate and efficient four-dimensional trajectory (4DT) calculation is crucial for Trajectory Based Operation (TBO) of air transportation. In order to create precise 4DT and provide a reference for the fine-grained Air Traffic Management(ATM), an integrated trajectory clustering for trajectory generation framework is proposed mining the main trajectory patterns and realizing the large-scale 4DT generation fast. Firstly, DBSCAN and TraClus clustering methods are accommodated into one integrated procedure to extract both the global and local features of trajectory. Then, some typical trajectory points mined above are adopted as the origin and destination waypoints to generate the target air-route trajectory via a directed network with transition probability. Two trajectory generation strategies are presented further to obtain the most possible trajectory and its related feasible trajectory set. Finally, through experiments with trajectory data collected on air-route from Beijing to Guangzhou, China, both the validity and accuracy of the proposed method are verified. © 2022 IEEE.

关键词： Trajectories

来源：评论

学校读者我要写书评

暂无评论

A Learning Estimation Approach for Arrival and Departure Capacity considering Weather Impact 41

A Learning Estimation Approach for Arrival and Departure Cap...

引用

41st IEEE/AIAA Digital Avionics Systems Conference, DASC 2022

作者： Tang, Shuo Fang, Quan Yang, Yang Chen, Jiatong Cai, Kaiquan Beihang University School of Electronic and Information Engineering Beijing China Institute of Automation Cas National Lab of Pattern Recognition Beijing China Beihang University Research Institute for Frontier Science Beijing China

ISBN: (数字)9781665486071

ISBN: (纸本)9781665486071

During severe weather, traffic managers in all facilities of terminal area are busy managing the air traffic within their airspace. In this case, shared capacity information is expected to assist decision making thus better coordinating the arrival and departure operation. Therefore, an integrated Estimation of the Arrival and Departure Capacity under Weather Impact is a crucial task to improve the operational safety and efficiency level in terminal airspace. However, currently, capacity estimation relies heavily on controller's empirical judgment and existing learning-based methods have limited application scenarios as they ignore the temporal and spatial effects of weather. This paper presents a Spatio-Temporal Graph Convolutional Network with Self-Enhancement mechanism and Initial residual (STGCNSEI) estimation model that extracts temporal characteristics and topological influences from coarse estimates to finer ones. Firstly, the XGBoost regression and the max-flow min-cut approach are adopted to preliminarily obtain a rough estimation of airport and fixes capacities respectively. Then, the STGCNSEI model is applied to further integrate the detailed temporal and spatial features in order to elevate estimation precision. Finally, the feasibility and accuracy of the proposed method is validated via a case study of Chengdu Terminal airspace, China. Results demonstrate that the proposed STGCNSEI model has a promising application prospect. © 2022 IEEE.

关键词： Decision making

来源：评论

学校读者我要写书评

暂无评论

Achievement Degree Evaluation Method of Operating System Course Targets under the Background of Engineering Education Accreditation 3

Achievement Degree Evaluation Method of Operating System Cou...

引用

3rd IEEE International Conference on Computer Science and Educational Informatization, CSEI 2021

作者： Zhang, Fan Zhang, Xinhong Henan University School of Computer and Information Engineering Kaifeng China Institute of Image Processing and Pattern Recognition Henan University Kaifeng China

ISBN: (纸本)9781728184739

Engineering education accreditation emphasizes the cultivation of students' ability to solve complex engineering problems, including the cultivation of students' ability to analyze, design and develop solutions, research, application of modern tools and other qualities related to non-technical factors to solve complex engineering problems. Operating system course is the core course of computer science and technology major in colleges and universities. The understanding and mastery of operating system course directly affects students' computer professional level. In view of the new requirements of new engineering construction and engineering education accreditation for major curriculum construction and personnel training, as well as the current characteristics of software engineering and personnel training, this paper, based on the core concept of OBE and complex engineering problems of engineering education accreditation, evaluates the achievement of operating system curriculum requirements under the background of engineering education accreditation. This paper analyzes the effectiveness and puts forward a new method of major curriculum construction and evaluation to meet the requirements of new engineering construction and engineering education accreditation. © 2021 IEEE.

关键词： Engineering education

来源：评论

学校读者我要写书评

暂无评论

Investigation of the J-TEXT plasma events by k-means clustering algorithm

引用

Plasma Science and Technology 2023年第8期25卷 38-43页

作者：李建超张晓卿张昱 Abba Alhaji BALA 柳惠平周帼红王能超李达陈忠勇杨州军陈志鹏董蛟龙丁永华 the J-TEXT Team Hubei Key Laboratory of Optical Information and Pattern Recognition Wuhan Institute of TechnologyWuhan 430205People's Republic of China International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics State Key Laboratory of Advanced Electromagnetic Engineering and TechnologySchool of Electrical and Electronic EngineeringHuazhong University of Science and TechnologyWuhan 430074People's Republic of China Department of Physics Federal University DutseJigawa 720101Nigeria 不详

Various types of plasma events emerge in specific parameter ranges and exhibit similar characteristics in diagnostic signals,which can be applied to identify these events.A semisupervised machine learning algorithm,the k-means clustering algorithm,is utilized to investigate and identify plasma events in the J-TEXT *** method can cluster diverse plasma events with homogeneous features,and then these events can be identified if given few manually labeled examples based on physical understanding.A survey of clustered events reveals that the k-means algorithm can make plasma events(rotating tearing mode,sawtooth oscillations,and locked mode)gathering in Euclidean space composed of multi-dimensional diagnostic data,like soft x-ray emission intensity,edge toroidal rotation velocity,the Mirnov signal amplitude and so *** on the cluster analysis results,an approximate analytical model is proposed to rapidly identify plasma events in the J-TEXT *** cluster analysis method is conducive to data markers of massive diagnostic data.

关键词： k-means cluster analysis plasma event machine learning

来源：评论

学校读者我要写书评

暂无评论

Vehicle Routing Problem with Time Windows using Hybrid Metaheuristic Dragonfly Algorithm and Variable Neighborhood Search: Work on Progress 11

Vehicle Routing Problem with Time Windows using Hybrid Metah...

引用

11th International Conference on Electrical Engineering, Computer Science and Informatics, EECSI 2024

作者： Yunita, Yunita Stiawan, Deris Rini, Dian Palupi Sriwijaya University Faculty of Computer Science Indonesia Sriwijaya University Intelligent System Research Group Faculty of Computer Science South Sumatera Palembang Indonesia Sriwijaya University Communication Network and Information Security Research Group Faculty of Computer Science South Sumatera Palembang Indonesia Sriwijaya University Image Processing Dan Pattern Recognition Laboratory Group Faculty of Computer Science South Sumatera Palembang Indonesia

ISBN: (纸本)9798350355314

One of the problems with Smart Transportation is the problem of cost and travel time. This problem is known as the Variable Routing Problem (VRP). In some real cases, in addition to considering route selection, there are limitations on the capacity and time period for the vehicle to serve each customer known as Time Windows, so VRP has developed into Variable Routing Problem with Time Windows (VRPTW). One way to solve VRPTW is to use metaheuristic methods. However, the metaheuristic method has a weakness, which are it can get stuck in local optimums and fail to find better global solutions. To overcome this, a combination with other methods or known as hybrid is needed. The formulation of this research problem is how to develop a Smart transportation model using a metaheuristic hybrid algorithm. The purpose of this study is to develop a Smart transportation model using a metaheuristic hybrid algorithm. The method used is to combine two metaheuristic algorithms, which are the Dragonfly Algorithm (DA) and the Variable Neighborhood Search (VNS). The result of this research is a new algorithm model which is a hybrid between DA and VNS. Through a combination of global exploration using DA and local exploitation using VNS, this algorithm is expected to be able to find a better solution and faster convergence in solving VRPTW problems © 2024 IEEE.

关键词： Travel time

来源：评论

学校读者我要写书评

暂无评论

Decentralized Kernel Ridge Regression Based on Data-Dependent Random Feature

arXiv

引用

arXiv 2024年

作者： Yang, Ruikai He, Fan He, Mingzhen Yang, Jie Huang, Xiaolin The Institute of Image Processing and Pattern Recognition The MOE Key Laboratory of System Control and Information Processing Shanghai Jiao Tong University Shanghai200240 China The STADIUS Center for Dynamical Systems Signal Processing and Data Analytics KU Leuven LeuvenB-3001 Belgium

Random feature (RF) has been widely used for node consistency in decentralized kernel ridge regression (KRR). Currently, the consistency is guaranteed by imposing constraints on coefficients of features, necessitating that the random features on different nodes are identical. However, in many applications, data on different nodes varies significantly on the number or distribution, which calls for adaptive and data-dependent methods that generate different RFs. To tackle the essential difficulty, we propose a new decentralized KRR algorithm that pursues consensus on decision functions, which allows great flexibility and well adapts data on nodes. The convergence is rigorously given and the effectiveness is numerically verified: by capturing the characteristics of the data on each node, while maintaining the same communication costs as other methods, we achieved an average regression accuracy improvement of 25.5% across six real-world data sets. Copyright © 2024, The Authors. All rights reserved.

关键词： Consensus algorithm

来源：评论

学校读者我要写书评

暂无评论

Simultaneously Infer Cell Pseudotime,Velocity Field and Gene Interaction from Multi-Branch scRNA-seq Data with scPN

arXiv

引用

arXiv 2024年

作者： Zhou, Zhen Li, Jiachen Xin, Hongyi Pan, Xiaoyong Shen, Hong-Bin Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai200240 China Global Institute of Future Technology Shanghai Jiao Tong University Shanghai200240 China

Modeling cellular dynamics from single-cell RNA sequencing (scRNA-seq) data is critical for understanding cell development and underlying gene regulatory relationships. Many current methods rely on single-cell velocity to obtain pseudotime, which can lead to inconsistencies between pseudotime and velocity. It is challenging to simultaneously infer cell pseudotime and gene interaction networks, especially in multi-branch differentiation scenarios. We present single-cell Piecewise Network (scPN), a novel high-dimensional dynamical modeling approach that iteratively extracts temporal patterns and inter-gene relationships from scRNA-seq data. To tackle multi-branch differentiation challenges, scPN models gene regulatory dynamics using piecewise gene-gene interaction networks, offering an interpretable framework for deciphering complex gene regulation patterns over time. Results on synthetic data and multiple scRNA-seq datasets demonstrate the superior performance of scPN in reconstructing cellular dynamics and identifying key transcription factors involved in development compared to existing methods. To the best of our knowledge, scPN is the first attempt at modeling that can recover pseudotime, velocity fields, and gene interactions all at once on multi-branch datasets. Copyright © 2024, The Authors. All rights reserved.

关键词： Transcription

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：