检索结果-内蒙古大学图书馆

4th IEEE International Conference on Pattern Recognition and Machine Learning, PRML 2023

作者： Ahat, Arzigul Firkat, Eksan Mijit, Tayir Hamdulla, Askar Xinjiang University of China Xinjiang Key Laboratory of Signal Detection and Processing School of Information Science and Engineering Urumqi China

ISBN: (纸本)9798350324303

Mapping off-road terrain is a challenging task, especially when compared to urban terrain. The complexity and roughness of off-road terrain make it difficult to achieve accurate mapping results. Obstacle detection in such environments is also a challenging task, as the performance of LiDAR-based methods depends on the point cloud density at the location of the obstacle. High line beam LiDAR is expensive, while low line beam LiDAR is sparse, making it difficult to obtain dense point clouds. In this paper, we propose a multi-LiDAR fusion approach that uses roof LiDAR as the primary sensor and front LiDAR as a supplement. The approach employs point cloud filtering, dedistortion, and point cloud registration to fuse data from multiple sensors. This results in a richer view of the environment and more intensive measurement data. We combine the voxel map with a twodimensional grid map and accelerate the voxel map using a GPU to generate accurate maps. Experimental results show that our proposed method achieves the highest accuracy and recall rates, improving the accuracy and robustness of obstacle detection in off-road environments. © 2023 IEEE.

关键词： Optical radar

来源：评论

学校读者我要写书评

暂无评论

Deep neural network for MIMO-SCMA detection 5

Deep neural network for MIMO-SCMA detection

引用

2023 5th International Conference on Information Science, Electrical, and Automation Engineering, ISEAE 2023

作者： Zhang, Shiwei Ge, Wenping Key Laboratory of Signal Detection and Processing College of Information Science and Engineering Xinjiang University Xinjiang Urumqi830017 China

ISBN: (纸本)9781510667440

This article introduces deep learning into the multiple-input multiple-output (MIMO) sparse code multiple access (SCMA) system and proposes a MIMO-SCMA detection scheme based on deep neural networks (DNN) to improve bit error rate (BER) performance. The DNN learns the codebook of each user through channel feature learning on different transmission antennas. The fully connected DNN is designed as the decoder at the receiving end, which does not require traditional multi-antenna detection and multi-user detection, and can obtain user data with one decoding operation. The encoder and decoder are trained using an end-to-end training method. All learning models of the DNN are generated offline and the learned models are used for online testing. In this model, the received signal and channel coefficients are set as input data, and the label corresponding to the transmitted symbol is set as output data for offline learning. After offline learning is completed, the model can be deployed online with fixed weights and biases. Through simulation experiments, the proposed DNN encoder-decoder method can reduce the BER and computational complexity of the receiver in the MIMO-SCMA system. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

A Lightweight Music Source Separation Model with Graph Convolution Network 1

引用

18th National Conference on Man-Machine Speech Communication, NCMMSC 2023

作者： Zhu, Mengying Wang, Liusong Hu, Ying School of Computer Science and Technology Xinjiang University Urumqi China Key Laboratory of Signal Detection and Processing in Xinjiang Urumqi China

ISBN: (数字)9789819706013

ISBN: (纸本)9789819706006

With the rapid advancement of deep neural networks, there has been a significant improvement in the performance of music source separation methods. However, most of them primarily focus on improving their separation performance, while ignoring the issue of model size in the real-world environments. For the application in the real-world environments, in this paper, we propose a lightweight network combined with the Graph convolutional network Attention (GCN_A) module for Music Source Separation (G-MSS), which includes an Encoder and four Decoders, each of them outputs a target music source. The G-MSS network adopts both time-domain and frequency-domain L1 losses. The ablation study verifies the effectiveness of our designed GCN Attention (GCN_A) module and multiple Decoders, and also make a visualization analysis of the main components in the G-MSS network. Comparing with the other 13 methods on the MUSDB18 dataset, our proposed G-MSS achieves comparable separation performance while maintaining the lower amount of parameters. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Enhancing Relational Classification Model Fusing Entity and Sentence Semantic Information 4

Enhancing Relational Classification Model Fusing Entity and ...

引用

4th IEEE International Conference on Pattern Recognition and Machine Learning, PRML 2023

作者： Wei, Chunji Qin, Xizhong Yuan, Taiping Xinjiang University of China Xinjiang Key Laboratory of Signal Detection and Processing School of Information Science and Engineering Urumqi830017 China

ISBN: (纸本)9798350324303

Relationship classification aims at mining the relationship between two entities in a sentence and is an essential basis for the information extraction task. However, traditional relationship classification methods, which only use the information of words in a sequence to predict the relationship between entities, cannot use the semantic information of entities and sentences for relationship classification. To solve these problems, this paper proposes a new model, ES-BERT: this model uses the word-to-word association in the BERT model to obtain the semantic information of entities by the positional encoding of entity pairs in a sentence. In addition, we use the rich semantic knowledge in the internal hidden layer of Bert to obtain the semantic information of the sentence, which can better help the model understand the sentence. Finally, the semantic information of entities and the semantic information of sentences in the hidden layer of BERT are fused, and the fused information is used for relationship classification. The experimental results show that the model proposed in this paper achieves an excellent F1 value of 88.87% on the SemEva1-2010 Task 8 dataset, which is better than other Bert-based Sota models. © 2023 IEEE.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

DFE-GCN: Dual Feature Enhanced Graph Convolutional Network for Controversy detection

引用

Computers, Materials & Continua 2023年第10期77卷 893-909页

作者： Chengfei Hua Wenzhong Yang Liejun Wang Fuyuan Wei KeZiErBieKe HaiLaTi Yuanyuan Liao College of Software Xinjiang UniversityUrumqi830000China Key Laboratory of Signal Detection and Processing in Xinjiang Uygur Autonomous Region Xinjiang UniversityUrumqi830000China Key Laboratory of Multilingual Information Technology in Xinjiang Uygur Autonomous Region Xinjiang UniversityUrumqi830000China

With the development of social media and the prevalence of mobile devices,an increasing number of people tend to use social media platforms to express their opinions and attitudes,leading to many online *** online controversies can severely threaten social stability,making automatic detection of controversies particularly *** controversy detection methods currently focus on mining features from text semantics and propagation ***,these methods have two drawbacks:1)limited ability to capture structural features and failure to learn deeper structural features,and 2)neglecting the influence of topic information and ineffective utilization of topic *** light of these phenomena,this paper proposes a social media controversy detection method called Dual Feature Enhanced Graph Convolutional Network(DFE-GCN).This method explores structural information at different scales from global and local perspectives to capture deeper structural features,enhancing the expressive power of structural ***,to strengthen the influence of topic information,this paper utilizes attention mechanisms to enhance topic features after each graph convolutional layer,effectively using topic *** validated our method on two different public datasets,and the experimental results demonstrate that our method achieves state-of-the-art performance compared to baseline *** the Weibo and Reddit datasets,the accuracy is improved by 5.92%and 3.32%,respectively,and the F1 score is improved by 1.99%and 2.17%,demonstrating the positive impact of enhanced structural features and topic features on controversy detection.

关键词： Controversy detection graph convolutional network feature enhancement social media

来源：评论

学校读者我要写书评

暂无评论

Sound source localization and detection based on parameter transfer learning 24

Sound source localization and detection based on parameter t...

引用

24th International Congress on Acoustics, ICA 2022

作者： Sun, Xinghao Ma, Mengzhen Hu, Ying Department of Information Science and Engineering Xinjiang University Urumqi830000 China Key Laboratory of Signal Detection and Processing in Xinjiang China

Sound source localization and detection is a joint task of identifying the presence of individual sound events and locating the sound sources in space. In order to promote the combination of two different tasks, we propose a sound source localization and detection method based on parameter transfer learning. Firstly, to solve the problem of uneven distribution of time-frequency dimension features, we propose a time-frequency dimension feature extractor, which extracts the features of time dimension and frequency dimension respectively and superimposes them together. Secondly, in order to promote feature interaction between different resolutions, we propose a refined feature learning module that can fuse features with different resolutions. Then, in order to reduce the training and reasoning time, we explore a temporal context representation module for learning temporal context information. The time context indicates that the module has better global feature capture capability and better parallelism capability compared with the recurrent neural network and gated recurrent unit. Finally, the rationality of our proposed model was verified by ablation experiments, and the effectiveness of our proposed model was verified by comparison with the best methods. © 2022 Proceedings of the International Congress on Acoustics. All rights reserved.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Improving Speech Perceptual Quality and Intelligibility Through Sub-band Temporal Envelope Characteristics 18th

Improving Speech Perceptual Quality and Intelligibility Thr...

引用

18th National Conference on Man-Machine Speech Communication, NCMMSC 2023

作者： Wu, Ruilin Huang, Zhihua Song, Jingyi Liang, Xiaoming School of Computer Science and Technology Xinjiang University Urumqi China Key Laboratory of Signal Detection and Processing in Xinjiang Urumqi China

ISBN: (纸本)9789819706006

In the speech enhancement (SE) model, using auxiliary loss based on acoustic parameters can improve enhancement effects. However, currently used acoustic parameters focus on frequency domain information, neglecting the importance of time domain information. This study explores the effects of training the SE model utilizing temporal envelope parameters as auxiliary losses on perceptual quality and intelligibility, primarily focusing on accurately extracting temporal envelope parameters and optimizing losses. Initially, the effectiveness and robustness of the temporal envelope parameters in optimizing the SE model were verified. Subsequently, a temporal envelope loss based on sub-band weighted (ENVLoss) is proposed based on the perceptual characteristics of human hearing. Then, a method for constructing a joint loss function is proposed, integrating temporal and spectral acoustic features to promote the enhancement effect. Experimental results show that temporal envelope characteristics improve both the time domain and the time-frequency domain SE models. Compared to other acoustic parameter losses, the SE model using a sub-band temporal envelope auxiliary loss shows improvement in the PESQ, STOI, and MOS estimation metrics. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Speech intelligibility

来源：评论

学校读者我要写书评

暂无评论

MSFSAN: A Novel Multi-Scale Spatio-Temporal Feature Screening Attention Network for Urban Carbon Emission Prediction

MSFSAN: A Novel Multi-Scale Spatio-Temporal Feature Screenin...

引用

2024 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2024

作者： Wang, Ben Qin, Xizhong Qin, Jiwei Zhang, Xiaoyu Ma, Haodong College of Computer Science and Technology Xinjiang University Xinjiang Key Laboratory of Signal Detection and Processing Ürümqi830046 China

ISBN: (纸本)9781665410205

In order to cope with the increasingly severe global energy conservation and emission reduction problems, research on urban carbon emission prediction is of great significance. The existing methods mainly use time series analysis to predict urban carbon emission, but there is a strong spatial correlation between the carbon emission data of several cities. Therefore, this paper designs a multi-scale spatial-temporal feature screening attention network to predict target cities' future carbon emission data. Firstly, this paper combines the daily carbon emission data of the near-neighbouring cities and the daily homologous emission data of the target city to analyze the urban carbon emission data from a spatio-temporal perspective. Then, this paper designs a multi-scale spatial interactive convolution module and a multi-scale temporal convolution module to extract multi-scale spatio-temporal features effectively. In addition, the feature screening module is designed to reduce the adverse effects of redundant features. Finally, the multi-scale features are used to predict the future carbon emissions of the target city through a predictor. The experimental results show that our prediction model is superior to the existing methods in six datasets. © 2024 IEEE.

关键词： Low emission

来源：评论

学校读者我要写书评

暂无评论

A Rumor detection Method Incorporating Correlation Features 3

A Rumor Detection Method Incorporating Correlation Features

引用

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Cai, Xiangsheng Tohti, Turdi Hamdulla, Askar Xinjiang University of China Xinjiang Key Laboratory of Signal Detection and Processing School of Information Science and Engineering Urumqi830017 China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

Most of the rumors on social media (e.g., Twitter) revolve around a certain domain or topic, which means that they want to express roughly the same meaning, only the way of expression or the words used are different. Most of the existing rumor detection work focuses on text semantic features, user information features, comment features, and communication structure features, but rarely utilizes the rumor features associated with the original text. Therefore, a GCN-BERT -based rumor detection method is proposed in this paper, which not only utilizes the rumor features of a message itself but also utilizes the rumor features of all the texts and words associated with it when performing rumor detection. Finally, three Twitter datasets are used for comparison experiments, and the results fully demonstrate that the method is more effective than existing rumor detection methods. © 2022 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

COT2DNet: Contextual 2D Attention for Scene Text Recognition 2

COT2DNet: Contextual 2D Attention for Scene Text Recognition

引用

2nd International Conference on Artificial Intelligence, Human-Computer Interaction and Robotics, AIHCIR 2023

作者： Qian, Yefei Ibrayim, Mayire Hamdulla, Askar Zhang, Chunhu Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China School of Computer Science and Technology Xinjiang University Xinjiang China

ISBN: (纸本)9798350360363

Although scene text recognition (STR) methods have made great progress, reading the text of irregularly shaped scenes remains a challenge. The current conversion of two-dimensional (2D) images to one-dimensional (1D) feature maps for sequence recognition may lose a large amount of spatial information and cannot recognize text with a large degree of curvature. In this paper, we propose an network infrastructure to recognize texts of arbitrary shapes, named COT2D Attention Text Recognition Network (COT2Net). Firstly, deformable convolution is added to the feature extraction module, which can perform feature extraction according to the shape of the character and expand the receptive field of feature extraction;Finally, in the Transformer encoding stage, the multi-head attention is changed to Contextual 2D Multi-head Attention (COT2D), which can make full use of the relationship between the input keys Q, K, and V to guide the learning of dynamic attention, thereby enhancing the visual expression ability of text images. The COT2DNet uses COT2D to learn the relevance of each character in a scene text image and achieves excellent results on public datasets. © 2023 IEEE.

关键词： Feature extraction

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：