检索结果-内蒙古大学图书馆

4th International Conference on Natural Language Processing, ICNLP 2022

作者： Ma, Sugang An, Wen Yang, Xiaobao Hou, Zhiqiang Xi'an University of Posts and Telecommunications Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an China Xi'an University of Posts and Telecommunications School of Computer Science and Technology Xi'an China Xi'an University of Posts and Telecommunications Xi'an Key Laboratory of Big Data and Intelligent Computing Xi'an China

ISBN: (数字)9781665495448

ISBN: (纸本)9781665495448

Most current object detection algorithms have the issue of missing objects due to occlusion. As the great difference of scale between occlusion objects and their integrity is affected, how to reduce the missing rate of occlusion objects has become a challenging problem in object detection algorithms. Aiming at the problem, this paper proposes an object detection algorithm based on multi-scale context information on the basis of the YOLOv4 algorithm framework. Firstly, the object features are extracted through the CSPDarkNet53 feature extraction network. Then, the proposed attention-guided multi-scale context information module (AMCM) enhances the ability to perceive object context information and endows features with different influence factors through attention weighting to improve feature discrimination. The experimental results show that the detection accuracy and recall of the proposed algorithm reach 83.2% and 96.4% respectively on PASCAL VOC dataset, and 83.2% and 91% on KITTI dataset, which improves the detection performance of YOLOv4 algorithm. © 2022 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

A Hybrid Prediction Method Based on MODWT-EMD for Time Series in Wireless Sensor networks 16

A Hybrid Prediction Method Based on MODWT-EMD for Time Serie...

引用

16th IEEE International Conference on Intelligent Systems and Knowledge Engineering, ISKE 2021

作者： Wang, Zhongmin Jia, Jingwen Gao, Cong Chen, Yanping Xia, Hong Xi'an University of Posts and Telecommunications Xi'an Key Laboratory of Big Data and Intelligent Computing Xi'an China Xi'an University of Posts and Telecommunications School of Computer Science Technology Xi'an China Xi'an University of Posts and Telecommunications Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an China

ISBN: (纸本)9781665405539

Time series is the main form of sensor data. Time series prediction is conducive to reduce the energy consumption of sensor nodes and increase the service life. As time cost and computation cost of the existing hybrid prediction methods for time series are not good, this paper proposes a hybrid prediction method based on MODWT-EMD for time series. Firstly, the original time series is decomposed into multiple frequency components by Maximal Overlapping Discrete Wavelet Transform (MODWT). Secondly, the multiple frequency components are decomposed successively by Empirical Mode Decomposition (EMD) to obtain IMF components and residuals. IMFs and residuals contain the feature information of time series. Then, the importance of the extracted feature information is scored and sorted by random forest. We select feature information with a higher score. Finally, the selected feature information is input into Bi-GRU for training. We predicted the original time series by the trained Bi-GRU and obtained the prediction results. The experimental results show that the method proposed in this paper can predict the time series effectively. © 2021 IEEE.

关键词： Empirical mode decomposition

来源：评论

学校读者我要写书评

暂无评论

Edge-cloud cooperation driven surface roughness classification method for selective laser melting

引用

Advanced Engineering Informatics 2025年 66卷

作者： Shuaiyin Ma Yuming Huang Yang Liu Zhiqiang Yan Jingxiang Lv Wei Cai School of Computer Science and Technology Xi’an University of Posts and Telecommunications Xi’an 710121 China Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi’an University of Posts and Telecommunications Xi’an 710121 China Xi’an Key Laboratory of Big Data and Intelligent Computing Xi’an 710121 China Shaanxi Union Research Centre of University and Enterprise for 5G+ Industrial Internet Communication Terminal Technology Xi’an University of Posts and Telecommunications Xi’an 710121 China Department of Management and Engineering Linköping University SE-581 83 Linköping Sweden Industrial Engineering and Management University of Oulu 90570 Oulu Finland Key Laboratory of Road Construction Technology and Equipment of MOE Chang’an University Xi’an 710064 China College of Engineering and Technology Southwest University Chongqing 400715 China

Additive manufacturing (AM) technology is extensively utilized in aerospace and industrial manufacturing. However, parts built using AM are susceptible to spheroidization, porosity, cracks, and poor surface quality, making it difficult to establish an actionable product quality degree. Hence, developing a reasonable method to equate product quality with a new degree and further analyzing the product quality based on these standards has proven effective for enhancing part quality in AM. To achieve this goal, this paper proposes a surface roughness classification method that utilizes surface roughness analysis and sample enhancement. This method leverages edge cloud cooperation to efficiently analyze and integrate data from different sensors, enabling real-time monitoring and adjustment of the manufacturing process. Subsequently, the quality degree analysis system was developed utilizing matter-element extension cloud model. Furthermore, a bidirectional-gated recurrent unit (Bi-GRU) based model for quality classification and recognition has been established, with Wasserstein generative adversarial network (WGAN) employed for sample enhancement to address the issue of imbalanced column classification and to enhance the accuracy of both classification and recognition. Finally, the results obtained from this case study demonstrate through comparative experiments that the proposed method for classifying surface roughness can accurately identify 98% of prepared samples.

关键词： Additive manufacturing Bi-GRU Clustering analysis Edge-cloud cooperation WGAN

来源：评论

学校读者我要写书评

暂无评论

A trajectory association strategy for multi-object tracking 6

A trajectory association strategy for multi-object tracking

引用

6th International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2023

作者： Ma, Sugang Duan, Shuaipeng Pu, Lei Hou, Zhiqiang School of Computer Science and Technology Xi'an University of Posts and Telecommunications Shaanxi Xi'an710121 China Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an University of Posts and Telecommunications Shaanxi Xi'an China School of Operational Support Rocket Force Engineering University Shaanxi Xi'an710025 China

ISBN: (纸本)9798400707674

When the multi-object tracking (MOT) algorithm confronts complex scenarios such as target occlusion and blurring, the trajectory missing and identity switching problems frequently occur. To address this issue, a trajectory association strategy for MOT is presented. First, using the similarity relationship between two frames, the similarity matrix is introduced to predict a more trustworthy offset vector and lower the number of Identity Switch. Then, a trajectory reconnection strategy is provided in order to retain a portion of the low-score detection frames due to occlusion and to perform secondary matching on the low-score detection frames in order to reduce missing trajectories. On a typical multi-object tracking dataset MOT17, experimental comparison study is performed with several known advanced algorithms, and MOTA and IDF1 achieve 67.9% and 67.8%, respectively. The experimental results show that the proposed algorithm better solves the trajectory missing and identity switching problems, and achieves excellent tracking performance in complex scenarios. © 2023 ACM.

关键词： Trajectories

来源：评论

学校读者我要写书评

暂无评论

Automatic ICD-10 Coding Based on Multi-Head Attention Mechanism and Gated Residual network

Automatic ICD-10 Coding Based on Multi-Head Attention Mechan...

引用

2021 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2021

作者： Wang, Xiaowei Han, Jungang Li, Ben Pan, Xiaoying Xu, Hui Xi'An University of Posts and Telecommunications School of Computer Science Xi'an China Shaanxi Key Laboratory of Network Data Analysis and Intelligence Processing Xi'an China Technology Co. LTD Beijing China

ISBN: (纸本)9781665401265

Classifying diseases in electronic medical records into corresponding ICD codes requires not only a large amount of medical knowledge but also a large number of coders, which is time-consuming and labor-consuming. Therefore, automatic coding is of great significance. This paper aims to build a deep learning model for automatic ICD-10 coding from a batch of Chinese electronic medical records. The data enhancement, convolutional neural network, attention mechanism, and the gating residual network proposed by the author were used to code ICD code corresponding to the distribution of medical record information by supervised learning. The benchmark model and ablation model were tested on a data set of Chinese electronic medical records. The effectiveness of the proposed modules, such as feature aggregation, multi-head attention mechanism, dilated convolution, and gating residuals, was verified. In the automatic ICD coding task for 104 diseases, the accuracy of the proposed method is 91.71%, and the F1-Score is 92.11%.1 © 2021 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

An Adaptive Model Update Object Tracking Algorithm based on DenseNet Features

An Adaptive Model Update Object Tracking Algorithm based on ...

引用

2020 International Conference on networking and network Applications, NaNA 2020

作者： Ma, Sugang Zhang, Lei Pu, Lei Yang, Xiaobao Hou, Zhiqiang Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an University of Posts and Telecommunications Xi'an shaanxi710121 China School of Computer Science and Technology Xi'an University of Posts and Telecommunications Xi'an shaanxi710121 China School of Information and Navigation Air Force Engineering University Xi'an shaanxi710077 China

ISBN: (纸本)9781728189543

In order to further improve the ability to deal with complex scenes, a visual tracking algorithm based on DenseNet features and model adaptive updating is proposed. Aiming to improve the feature representation ability of correlation filtering algorithm, a more discriminative DenseNet model is used as feature extractor. Meanwhile, in order to improve the long-term tracking performance, an adaptive model update strategy is proposed. The average peak correlation energy (APCE) based on response map is used to judge and evaluate the tracking result and the model is updated only when the confidence is high. The experimental results show that the proposed algorithm achieves higher success rate and accuracy compared to the other five related algorithms, and improves the robustness of object tracking in complex scenes such as occlusion and deformation. © 2020 IEEE.

关键词： Tracking (position)

来源：评论

学校读者我要写书评

暂无评论

Real-Time Semantic Segmentation Algorithm Based on Tversky Loss Function and Mixed Pooling

Real-Time Semantic Segmentation Algorithm Based on Tversky L...

引用

2023 International Conference on networking and network Applications, NaNA 2023

作者： Xu, Peng Zhao, Ziyi Ma, Sugang Xi'an Science and Technology Museum Shaanxi Xi'an710002 China School of Computer Science and Technology Xi'an University of Posts and Telecommunications Shaanxi Xi'an710121 China School of Computer Science and Technology Xi'an University of Posts and Telecommunications Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Shaanxi Xi'an710121 China

ISBN: (纸本)9798350327380

In recent years, semantic segmentation methods based on deep learning have made remarkable developments. Despite achieving high segmentation accuracy, the performance of real-time segmentation methods cannot satisfy real-world applications. In order to achieve a balance between segmentation accuracy and speed, a real-time semantic segmentation algorithm based on Tversky loss function and mixed pooling is proposed in this paper. A Short-Term Dense Concatenate network (STDC network) is used to construct the encoder, and a mixed pooling module is used for the final part of the encoder, which uses strip pooling and average pooling to enhance the feature representation while expanding the receptive field. Additionally, a Tversky-based loss function is used for the detail guidance module of the backbone network in the encoder, and a joint loss function is used to supervise the whole network's training. We achieved 76.9% mIoU at 110.7 FPS on the Cityscapes dataset, a 2.4% improvement in accuracy over the benchmark algorithm STDCSeg, and 72.2% mIoU at 177.6 FPS on the Camvid dataset, satisfied the requirements of the real-time segmentation task. © 2023 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Distance variety preserving hashing for large-scale retrieval

Distance variety preserving hashing for large-scale retrieva...

引用

作者： Zhai, Sheping School of Computer Science and Technology Xi'an University of Posts and Telecommunications Xi'an710121 China Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an University of Posts and Telecommunications Xi'an710121 China

Hashing techniques have been seen huge adoption in large-scale retrieval owing to the computational and storage efficiencies of binary codes. However, the binary codes might not preserve the semantic similarity powerfully and lead to poor accuracy. To evade binary limitation but remain hashing superiority, this paper introduces a novel asymmetric hashing framework for distance variety preserving, which yields preferable performance via real-valued query and multi-valued database with shorter code length. Concretely, a binary sparse representation is proposed for multivalued database encoding to make computation and storage complexity tolerable. By exploiting the supervised information, we learn a projective matrix that project data onto a sub-space in which the similarity between realvalued query and multi-valued database can be measured. Due to the discrete constraints, for different discrete variables, we present different meticulous discrete solutions for their sub-problems. Exhaustive experiments on three multilabel datasets demonstrate our approach outperforms most state-of-the-art hashing methods. © 2018 John Wiley & Sons, Ltd.

关键词： database systems

来源：评论

学校读者我要写书评

暂无评论

A Multi-stage Feature Enhancement network with Second-Attention for Image Captioning 6

A Multi-stage Feature Enhancement Network with Second-Attent...

引用

6th International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2023

作者： Yang, Xiaobao Yang, Yang Wu, Junsheng Ma, Sugang Hou, Zhiqiang Song, Bohui Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an University of Posts and Telecommunications Shaanxi Xi'an710121 China School of Computer Science and Technology Xi'an University of Posts and Telecommunications Shaanxi Xi'an710121 China School of Computer Science Northwestern Polytechnical University Xi'an710071 China

ISBN: (纸本)9798400707674

In the past few years, self-attention in Transformer has been widely used in natural language processing (NLP) and computer vision (CV) due to its excellent ability to capture global information, especially in image captioning, where the use of self-attention can significantly improve the representation of visual information. However, the obtained image feature from Transformer-based models suffers from two main problems, one is information redundancy due to the global aggregation of self-attention, and the other is the lack of semantic information caused by single-scale feature extraction. Therefore, we first propose second-attention in this paper, by redistributing image attention weights, second-attention can effectively remove irrelevant information and enhance the attention to essential objects and relations. Meanwhile, to enrich semantic information and further enhance the role of second-attention, we design a Multi-stage Feature Enhancement (MFE) network to improve the ability to represent visual information. After conducting extensive experiments on the MS COCO dataset, we achieve significant improvements in all popular benchmarks, particularly BLEU4 increased by 1.6%, CIDEr increased by 3.4%. © 2023 ACM.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Plagiarism Detection of Multi-threaded Programs Using Frequent Behavioral Pattern Mining

Plagiarism Detection of Multi-threaded Programs Using Freque...

引用

作者： Tian, Zhenzhou Wang, Qing Gao, Cong Chen, Lingwei Wu, DInghao School of Computer Science and Technology Xi'An University of Posts and Telecommunications China Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing Xi'an Shaanxi710121 China College of Information Sciences and Technology The Pennsylvania State University University ParkPA16802 United States

Software dynamic birthmark techniques construct birthmarks using the captured execution traces from running the programs, which serve as one of the most promising methods for obfuscation-resilient software plagiarism detection. However, due to the perturbation caused by non-deterministic thread scheduling in multi-threaded programs, such dynamic approaches optimized for sequential programs may suffer from the randomness in multi-threaded program plagiarism detection. In this paper, we propose a new dynamic thread-aware birthmark FPBirth to facilitate multi-threaded program plagiarism detection. We first explore dynamic monitoring to capture multiple execution traces with respect to system calls for each multi-threaded program under a specified input, and then leverage the Apriori algorithm to mine frequent patterns to formulate our dynamic birthmark, which can not only depict the program's behavioral semantics, but also resist the changes and perturbations over execution traces caused by the thread scheduling in multi-threaded programs. Using FPBirth, we design a multi-threaded program plagiarism detection system. The experimental results based on a public software plagiarism sample set demonstrate that the developed system integrating our proposed birthmark FPBirth copes better with multi-threaded plagiarism detection than alternative approaches. Compared against the dynamic birthmark System Call Short Sequence Birthmark (SCSSB), FPBirth achieves 12.4%, 4.1% and 7.9% performance improvements with respect to union of resilience and credibility (URC), F-Measure and matthews correlation coefficient (MCC) metric, respectively. © 2020 World Scientific Publishing Company.

关键词： Semantics

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：