检索结果-内蒙古大学图书馆

4th International Symposium on computer Engineering and Intelligent Communications, ISCEIC 2023

作者： Huang, Guimin Li, Ying Guo, Qingkai Deng, Nanxiao Lin, Jianxing Guilin University of Electronic Technology School of Computer Science and Information Security Guangxi Key Laboratory of Image and Graphic Intelligent Processing Guilin China

ISBN: (纸本)9798350306323

With online education rapidly developing, it is a significant issue to evaluate students' grasp of knowledge more accurately. Knowledge tracing models are good at it, in which convolutional knowledge tracing (CKT) based on convolutional neural networks(CNN) gets better performance. However, we found that the CKT model ignores the relation between the knowledge points, which we believe has an important impact on the correctness of students' answers. Therefore, we put forward a graph embedding based convolutional knowledge tracing model (CKT-GE), which feeds the relation between the knowledge points with other inputs into model for improving the effectiveness of knowledge tracing. Our model outperforms existing models on two classic datasets, and improves the performance over the original CKT model, allowing for more accurate assessment of student performance. © 2023 IEEE.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

Improving Speaker Verification Back-End with Graph Neural Networks

引用

Journal of Shanghai Jiaotong University (Science) 2025年 1-9页

作者： Chen, Jinfeng Fang, Zhihua He, Liang School of Computer Science and Technology Xinjiang University Urumqi830017 China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China School of Intelligence Science and Technology Xinjiang University Urumqi830017 China Department of Electronic Engineering Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China

Currently, research on speaker verification tasks is primarily concentrated on enhancing deep speaker models to extract high-quality speaker embeddings. Nevertheless, this speaker embeddings can be regarded as potential graph structures. Therefore, this paper introduces a graph neural network-based speaker verification back-end approach, treating the speaker embeddings derived from the front-end as graph structures and leveraging graph neural networks to uncover the relationships among embeddings, thereby achieving superior-quality speaker embeddings. In addition, we propose a group updating method to solve the problem of memory overflow when the number of nodes is excessive. Extensive experiments and ablation studies were carried out on the VoxCeleb dataset, and the experimental outcomes validate the effectiveness of our proposed graph neural network speaker back-end in significantly boosting the performance of speaker verification systems. (Figure presented.) © Shanghai Jiao Tong University 2025.

关键词： Graph embeddings

来源：评论

学校读者我要写书评

暂无评论

A Deep Knowledge Tracking Model Based on Forgetting Features and Self-Attention Mechanism 4

A Deep Knowledge Tracking Model Based on Forgetting Features...

引用

4th International Symposium on computer Engineering and Intelligent Communications, ISCEIC 2023

作者： Huang, Guimin Wei, Xiong Guo, Qingkai Deng, Nanxiao Lin, Jianxing Guilin University of Electronic Technology School of Computer Science and Information Security Guangxi Key Laboratory of Image and Graphic Intelligent Processing Guilin China

ISBN: (纸本)9798350306323

The integration of large education and artificial intelligence technologies is gradually deepening, and how to provide personalized user profiling services for learners is an important research problem. In response to the long-term dependency problem and lack of learning features of today's deep knowledge tracking models, a deep knowledge tracking model (FSA-MuLSTM) based on forgetting features and the self-attentiveness mechanism is proposed. Firstly, the model combines the learner's historical learning interaction tuple sequences with learning times and memory residuals to obtain the input vector based on the LSTM;secondly, the correlation weights between tuples are captured using the self-attention mechanism;finally, the deep information between tuple sequences is captured using the stacking of multilayer LSTMs. The probability of correctly answering the current question is derived from the hidden vectors and the self-attentive weights of the output of the multilayer LSTM. Two public datasets are used for the experiments, and the combined analysis outperforms existing knowledge-tracking models in terms of performance and improves the accuracy of prediction. © 2023 IEEE.

关键词： Multilayers

来源：评论

学校读者我要写书评

暂无评论

Anomalous Sound Detection Using Time-Frequency Feature and Mixbatch

引用

Journal of Shanghai Jiaotong University (Science) 2025年 1-8页

作者： Huang, Shun Zhang, Yunxiang Fang, Zhihua Tang, Minrui Xu, Ruifeng He, Liang School of Computer Science and Technology Xinjiang University Urumqi830017 China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China School of Intelligence Science and Technology Xinjiang University Urumqi830017 China Department of Electronic Engineering Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China

The sound emitted by machines under abnormal working conditions exhibits various frequency patterns. Currently, the most advanced anomalous sound detection (ASD) approach is to apply a multi-head self-attention mechanism to the Log-Mel spectrogram for automatic frequency pattern analysis. However, the Log-Mel spectrogram may filter out high-frequency components of abnormal sounds;thus the use of self-attention mechanisms on Log-Mel spectrogram seems to have certain limitations. In this paper, we construct a simple convolutional neural network to extract comprehensive frequency features from raw audio to complement spectral-temporal information fusion. The parameters of this neural network are continuously updated during the training process to extract better frequency features for downstream classification neural networks in ASD. Additionally, a method for data augmentation in the batch dimension was developed to help the classification model learn both types of comprehensive features simultaneously. The proposed method achieved an AUC of 94.13% and a pAUC of 89.09% on the DCASE 2020 Challenge Task 2 dataset. Even when using only the features proposed in this paper, an AUC of 84.62% was achieved. (Figure presented.). © Shanghai Jiao Tong University 2025.

关键词： Spectrographs

来源：评论

学校读者我要写书评

暂无评论

DensityLayout: Density-Conditioned Layout GAN for Visual-Textual Presentation Designs 12th

DensityLayout: Density-Conditioned Layout GAN for Visual-Te...

引用

12th International Conference on Image and Graphics, ICIG 2023

作者： Hsu, HsiaoYuan He, Xiangteng Peng, Yuxin Wangxuan Institute of Computer Technology Peking University Beijing China National Key Laboratory for Multimedia Information Processing Peking University Beijing China

ISBN: (纸本)9783031463075

Generating layouts for visual-textual presentation designs aims at arranging elements such as logo, text, and underlay on the given images, which is the key to automating poster designs. It is challenging since the compositions of images, the spatial patterns of layout elements, and their cross-relationships need to be simultaneously considered. Existing works focus on the cross-relationships and either (1) suffer from the instability of adopting off-the-shelf saliency maps as prior knowledge or (2) require the semantic content of each element. To this end, this paper presents an efficient density paradigm that requires neither off-the-shelf models nor additional training data other than image-layout pairs. Under this paradigm, a three-stage approach using GAN is proposed, entitled DensityLayout. First, a density mapping network weakly supervised by the custom consistency loss will translate given images to spatial distributions of elements. Second, a multi-scale strategy is proposed to enhance understanding of the maps, and a generator conditioned on these visual features will generate preliminary layouts. Finally, a directed graph representation illustrating the inclusion relationships between elements is presented, and a graph convolution network will fine-tune the layouts. The effectiveness of the proposed approach is validated on CGL-Dataset, showing it achieves the best performance by generating visually appealing layouts for visual-textual presentation designs of diverse images. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Directed graphs

来源：评论

学校读者我要写书评

暂无评论

Non-intrusive load monitoring based on a combination of Transformer and CNN 4

Non-intrusive load monitoring based on a combination of Tran...

引用

4th International Conference on Energy, Power and Electrical Engineering, EPEE 2024

作者： Cheng, Yiqi Zhong, Yanru Guilin University of Electronic Technology College of Computer and Information Security Guangxi Guilin China Guilin University of Electronic Technology Guangxi Key Laboratory of Image And Graphic Intelligent Processing Guangxi Guilin China

ISBN: (纸本)9798331518066

Non-intrusive load monitoring (NILM) estimates the energy usage of individual devices by analyzing the total power meter data from a household. To address the issue of decreased monitoring accuracy with an increasing number of electricity meters, this paper proposes a solution for single appliance load monitoring (Tra-CNN). The solution uses the UK-DALE dataset, which contains high and low frequencies electricity consumption data of multiple households in the UK and the electricity consumption data of each household appliance, and uses the active power of the appliance as the input of the algorithm. The first part of the model consists of a convolutional layer, a normalization layer, and a dropout layer, with the primary task of capturing local dependencies in the data. The second part consists of a Transformer, whose self-attention mechanism associates features at different positions in the sequence to achieve modeling of long-term dependencies. The third part introduces a convolutional neural network, which uses a sliding window for convolution operations to extract local features from different positions and further extract features from sequence data. The final decoder consists of a transposed convolutional layer and a normalization layer, designed to revert the learned features to their original dimensions. Through experimental verification, the effectiveness and superiority of Tra-CNN are shown, which provides important ideas and methods for the further development of NILM technology. © 2024 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Smart Power Safety Hazard Inspection System Based on YOLOv7 1st

Smart Power Safety Hazard Inspection System Based on YOLOv7

引用

1st International Conference on Applied Intelligence, ICAI 2023

作者： Liang, Yiheng Li, Xiaoming Deng, Zhenrong School of Computer and Information Security Guilin University of Electronic Technology Guilin541004 China Guangxi Key Laboratory of Image and Graphic Intelligent Processing Guilin University of Electronic Technology Guilin541004 China

ISBN: (纸本)9789819708260

Power safety is closely related to people’s well-being. Electric power sectors regularly inspect and maintain power lines to guarantee people’s safe and stable use of electricity, and the current mainstream inspection method is to use UAVs for power inspection. Traditional inspection and detection methods have many shortcomings, such as high labor costs, slow inspection speed, low detection efficiency, and single detection targets. In this paper, a set of smart power inspection systems based on YOLOv7 was designed to check the common potential security risks of power lines such as bird nests, insulators, garbage, and hardware. After the detection, the detection report could be generated. The system adopted the YOLOv7 target detection algorithm. The accuracy rate of all detection targets was up to 93.9%, and the mAP @.5% reached 95.9%. Therefore, the algorithm could greatly improve the detection efficiency and greatly facilitate the electric power sectors to check power lines, so it has a high use value. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Inspection

来源：评论

学校读者我要写书评

暂无评论

RGB-guided hyperspectral image super-resolution with deep progressive learning

引用

CAAI Transactions on Intelligence technology 2024年第3期9卷 679-694页

作者： Tao Zhang Ying Fu Liwei Huang Siyuan Li Shaodi You Chenggang Yan School of Computer Science and Technology Beijing Institute of TechnologyBeijingChina Satellite Information Intelligent Processing and Application Research Laboratory Beijing Institute of Remote SensingBeijingChina Key Laboratory of Spectral Imaging Technology Xi'an Institute of Optics and Precision Mechanics of Chinese Academy of SciencesXi'anChina Institute of Informatics University of AmsterdamAmsterdamNetherlands School of Communication Engineering Hangzhou Dianzi UniversityHangzhouChina

Due to hardware limitations,existing hyperspectral(HS)camera often suffer from low spatial/temporal ***,it has been prevalent to super-resolve a low reso-lution(LR)HS image into a high resolution(HR)HS image with a HR RGB(or mul-tispectral)image *** approaches for this guided super-resolution task often model the intrinsic characteristic of the desired HR HS image using hand-crafted ***,researchers pay more attention to deep learning methods with direct supervised or unsupervised learning,which exploit deep prior only from training dataset or testing *** this article,an efficient convolutional neural network-based method is presented to progressively super-resolve HS image with RGB image ***-ically,a progressive HS image super-resolution network is proposed,which progressively super-resolve the LR HS image with pixel shuffled HR RGB image ***,the super-resolution network is progressively trained with supervised pre-training and un-supervised adaption,where supervised pre-training learns the general prior on training data and unsupervised adaptation generalises the general prior to specific prior for variant testing *** proposed method can effectively exploit prior from training dataset and testing HS and RGB images with spectral-spatial *** has a good general-isation capability,especially for blind HS image *** experimental results show that the proposed deep progressive learning method out-performs the existing state-of-the-art methods for HS image super-resolution in non-blind and blind cases.

关键词： computer vision deep neural networks image processing image resolution unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

A Compact Circularly Polarized Antenna for Mobile Satellite Communication Applications 12

A Compact Circularly Polarized Antenna for Mobile Satellite ...

引用

12th IEEE Asia-Pacific Conference on Antennas and Propagation, APCAP 2024

作者： Liu, Yi-Ning Lu, Kai Zheng, Bin Zeng, Lingxin Yang, Nan Sun Yat-sen University School of Electronics and Information Technology Guangdong Provincial Key Laboratory of Optoelectronic Information Processing Chips and Systems Guangzhou510006 China China Satellite Network Exploration Co. Ltd Chongqing401100 China

ISBN: (纸本)9798350351019

A compact circularly polarized (CP) substrate-based antenna for mobile satellite service (MSS) is presented. The substrate has a metal ground fully covering its bottom, and an L-shaped shorted strip is loaded on its top. The shorting pin, close to the feeding probe, significantly helps reduce the antenna size. Moreover, the top strip is bent into an L shape to obtain good CP fields in the far field. Within a space of 0.15 λ0 × 0.1 λ0 × 0.03 λ0, the proposed antenna exhibits a -4-dB gain bandwidth of 21 MHz(1510-1531 MHz), which can fully accommodate the 7-MHz L-band MSS requirements. © 2024 IEEE.

关键词： Circular polarization

来源：评论

学校读者我要写书评

暂无评论

An Analysis Model of English Text Coherence Based on RST Dependency Relationship 4

An Analysis Model of English Text Coherence Based on RST Dep...

引用

4th International Conference on information Science, Parallel and Distributed Systems, ISPDS 2023

作者： Huang, Guimin Shuai, Wanying Guo, Qingkai Deng, Nanxiao Lin, Jianxing School of Computer Science and Information Security Guilin University of Electronic Technology Guangxi Key Laboratory of Image and Graphic Intelligent Processing Guilin China

ISBN: (纸本)9798350337181

Text coherence analysis is an important and challenging task that is essential for subtasks such as automatic summarisation, viewpoint extraction and machine translation in natural language processing (NLP). A large body of previous work has used linguistic features such as lexical, syntactic and entity features to capture relatively shallow coherence features, while ignoring deeper logically relevant features hidden in the text. We propose an end-to-end English text coherence analysis model (RGCM for short) that incorporates RST and graph convolutional neural networks (GCN). The intra- and inter-sentence logical relations of the input text are first mapped onto a discourse relationship tree, and then transformed into an RST-dependent context graph by certain pruning strategies. Subsequently, we propose a GCN-based text coherence assessment framework to capture intra- and inter-sentence interactions to assess text coherence. We conducted experiments on two different coherence assessment datasets and achieved accuracy rates of 95.5% and 97.8% respectively. © 2023 IEEE.

关键词： Convolutional neural networks

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：