检索结果-内蒙古大学图书馆

19th International Conference on Advanced Data Mining and Applications, ADMA 2023

作者： Liang, Wei Zhang, Kai Cao, Peng Liu, Xiaoli Yang, Jinzhu Zaiane, Osmar R. Computer Science and Engineering Northeastern University Shenyang China Key Laboratory of Intelligent Computing in Medical Image of Ministry of Education Northeastern University Shenyang China National Frontiers Science Center for Industrial Intelligence and Systems Optimization Shenyang110819 China DAMO Academy Alibaba Group Hangzhou China Alberta Machine Intelligence Institute University of Alberta EdmontonAB Canada

ISBN: (纸本)9783031466700

Alzheimer’s disease (AD) is a common chronic neurodegenerative disease and the accurate prediction of the clinical cognitive performance is important for diagnosis and treatment. Recently, multi-task feature learning (MTFL) methods with sparsity-inducing regularization have been widely investigated on cognitive performance prediction tasks. Although they have proved to achieve improved performance compared with single-task learning, the major challenges are still not fully resolved. They involve how to capture the non-linear correlation among the tasks or features, and how to introduce the learned correlation for guiding the MTFL learning. To resolve these challenges, we introduce a correlation structure learning method through self-attention learning and sequence learning for jointly capturing the complicated but more flexible relationship for features and tasks, respectively. Moreover, we develop a dual graph regularization to encode the inherent correlation and an efficient optimization algorithm for solving the nonsmooth objective function. Extensive results on the ADNI dataset demonstrate that the proposed joint training framework outperforms existing methods and achieves state-of-the-art prediction performance of AD. Specifically, the proposed algorithm achieves an nMSE (normalized Mean Squared Error)/wR (weighted R-value) of 3.808/0.438, obtaining a relative improvement of 3.84 % /7.35 % compared with the MTFL method. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Neurodegenerative diseases

来源：评论

学校读者我要写书评

暂无评论

A Comprehensive Survey of 3D Dense Captioning: Localizing and Describing Objects in 3D Scenes

arXiv

引用

arXiv 2024年

作者： Yu, Ting Lin, Xiaojun Wang, Shuhui Sheng, Weiguo Huang, Qingming Yu, Jun School of Information Science and Technology Hangzhou Normal University Hangzhou311121 China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China School of Computer Science and Technology University of Chinese Academy of Sciences Beijing101408 China Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University Hangzhou310018 China

Three-Dimensional (3D) dense captioning is an emerging vision-language bridging task that aims to generate multiple detailed and accurate descriptions for 3D scenes. It presents significant potential and challenges due to its closer representation of the real world compared to 2D visual captioning, as well as complexities in data collection and processing of 3D point cloud sources. Despite the popularity and success of existing methods, there is a lack of comprehensive surveys summarizing the advancements in this field, which hinders its progress. In this paper, we provide a comprehensive review of 3D dense captioning, covering task definition, architecture classification, dataset analysis, evaluation metrics, and in-depth prosperity discussions. Based on a synthesis of previous literature, we refine a standard pipeline that serves as a common paradigm for existing methods. We also introduce a clear taxonomy of existing models, summarize technologies involved in different modules, and conduct detailed experiment analysis. Instead of a chronological order introduction, we categorize the methods into different classes to facilitate exploration and analysis of the differences and connections among existing techniques. We also provide a reading guideline to assist readers with different backgrounds and purposes in reading efficiently. Furthermore, we propose a series of promising future directions for 3D dense captioning by identifying challenges and aligning them with the development of related tasks, offering valuable insights and inspiring future research in this field. Our aim is to provide a comprehensive understanding of 3D dense captioning, foster further investigations, and contribute to the development of novel applications in multimedia and related domains. Copyright © 2024, The Authors. All rights reserved.

关键词： Data handling

来源：评论

学校读者我要写书评

暂无评论

Prompting Video-Language Foundation Models with Domain-specific Fine-grained Heuristics for Video Question Answering

arXiv

引用

arXiv 2024年

作者： Yu, Ting Fu, Kunhao Wang, Shuhui Huang, Qingming Yu, Jun School of Information Science and Technology Hangzhou Normal University Hangzhou311121 China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China School of Computer Science and Technology University of Chinese Academy of Sciences Beijing101408 China Key Laboratory of Complex Systems Modeling and Simulation School of Computer Science and Technology Hangzhou Dianzi University Hangzhou310018 China

Video Question Answering (VideoQA) represents a crucial intersection between video understanding and language processing, requiring both discriminative unimodal comprehension and sophisticated cross-modal interaction for accurate inference. Despite advancements in multi-modal pre-trained models and video-language foundation models, these systems often struggle with domain-specific VideoQA due to their generalized pre-training objectives. Addressing this gap necessitates bridging the divide between broad cross-modal knowledge and the specific inference demands of VideoQA tasks. To this end, we introduce HeurVidQA, a framework that leverages domain-specific entity-action heuristics to refine pre-trained video-language foundation models. Our approach treats these models as implicit knowledge engines, employing domain-specific entity-action prompters to direct the model’s focus toward precise cues that enhance reasoning. By delivering fine-grained heuristics, we improve the model’s ability to identify and interpret key entities and actions, thereby enhancing its reasoning capabilities. Extensive evaluations across multiple VideoQA datasets demonstrate that our method significantly outperforms existing models, underscoring the importance of integrating domain-specific knowledge into video-language models for more accurate and context-aware VideoQA. Copyright © 2024, The Authors. All rights reserved.

关键词： Inference engines

来源：评论

学校读者我要写书评

暂无评论

Undersampling Ensemble and Deep Learning - Based Landslide Susceptibility Mapping Method for Geological Hazard Warning

Undersampling Ensemble and Deep Learning - Based Landslide S...

引用

Chinese Automation Congress (CAC)

作者： Yulong Fang Chao Gan Weihua Cao School of Automation China University of Geosciences Wuhan China Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Wuhan China Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China

Landslide is the most frequent geological hazard. Landslide susceptibility mapping (LSM) can be used to predict the possibility of landslide occurring at a certain location. In this paper, an undersampling ensemble and deep learning - based landslide susceptibility mapping method for geological hazard warning is proposed. The Zigui to Badong section of the Three Gorges Reservoir is taken as the study area. Firstly, the correlation analysis of fourteen landslide influencing factors is carried out and two unimportant factors are eliminated. Then, an EasyEnsemble - one dimensional convolutional neural network (EE-1DCNN) model is constructed with the remaining twelve factors as inputs. Finally, the proposed EE-1DCNN model is compared with two well-known methods on test data, and a landslide susceptibility map of the study area is obtained based on the EE-1DCNN model. As the experimental result shows, the proposed EE-1DCNN model achieves superior AUC, accuracy and recall of 0.909, 91.7% and 85.1%. The applicability of the proposed method is proved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Novel intelligent Method for Extracting the Multi-Domain Features of Ground Penetrating Radar Data

A Novel Intelligent Method for Extracting the Multi-Domain F...

引用

Chinese Automation Congress (CAC)

作者： Run Zhang Chao Gan Weihua Cao School of Automation China University of Geosciences Wuhan China Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Wuhan China Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China

Ground Penetrating Radar (GPR) features are vital for accurately predicting the formation environment in the tunnel engineering. In this paper, a novel intelligent method for extracting the multi-domain (time domain, time frequency domain, and spatial domain) features of ground penetrating radar data is proposed. Firstly, the GPR exploration data is read to obtain the relative amplitude matrix. Secondly, multi-domain features are obtained using the following methods. The relative amplitude matrix of GPR is averaged by rows to obtain the time domain feature called the average relative amplitude(ARA). The S-transform is used to extract the time frequency domain feature called the average maximum weight frequency(AMWF) of the electromagnetic waves. And the events of the GPR images are highlighted by using Gaussian filtering and edge detection, and the spatial domain feature called the maximum event length(M EL) is obtained through contour detection. Finally, the three extracted multi-domain features are stored in the GPR feature database. Compared simulation results verify the effectiveness of the proposed method.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Bounded Tracking of Uncertain Multiagent systems with Time-Delays and Multiplicative Noises

Bounded Tracking of Uncertain Multiagent Systems with Time-D...

引用

Industrial Cyber-Physical systems (ICPS)

作者： Guoliu Chen Xiaofeng Zong School of Automation China University of Geosciences Wuhan China Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Wuhan China Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China

This paper focuses on the bounded tracking control of general linear multi-agent systems(MASs), considering the effects of inevitable communication time-delays, measurement noises, and uncertain disturbances in practical applications. Firstly, the bounded tracking control problem of uncertain MASs under multiplicative noises is transformed into the boundedness problem of stochastic differential delay equations. Then, the upper bound of agent tracking is calculated by means of linear variation, variation of constants formula and stochastic analysis theory, and sufficient conditions are given for the system to achieve the bounded tracking control.

关键词： Sufficient conditions Upper bound Uncertainty Measurement uncertainty Stochastic processes Cyber-physical systems Mathematical models

来源：评论

学校读者我要写书评

暂无评论

DAANet: Dual Attention Aggregating Network for Salient Object Detection

DAANet: Dual Attention Aggregating Network for Salient Objec...

引用

2023 IEEE International Conference on Robotics and Biomimetics, ROBIO 2023

作者： Li, Yijie Wang, Hewei Li, Zhenqi Wang, Shaofan Dev, Soumyabrata Zuo, Guoyu Beijing University of Technology Beijing-Dublin International College Beijing100124 China Beijing University of Technology Faculty of Information Technology Beijing100124 China University College Dublin School of Computer Science Dublin Ireland Beijing University of Technology Faculty of Information Technology China Beijing Key Laboratory of Computing Intelligence and Intelligent Systems Beijing100124 China

ISBN: (纸本)9798350325706

When the Convolutional neural network (CNN) has been introduced for computer vision, many researches use CNN-based model to perform salient object detection (SOD). In recent years, The feature pyramid network (FPN) based structure is more popular for salient object detection tasks. In this paper, to improve the overall performance of salient object detection tasks, we propose a dual attention aggregating network (DAANet), which is an FPN-based deep convolutional neural network with a dual attention aggregation module (DAAM) and dilated refinement block (DRB). The DRB module uses convolutions with different dilation rates to expand the receptive field. The DAAM considers the salient map prediction from the low-level output as pseudo-attention which can efficiently aggregate multi-scale information. The convolution block attention module (CBAM) in DAAM can refine the aggregation of pseudo-attention which enables better performance. We evaluate DAANet on six benchmark datasets that prove the effectiveness of DAANet and its components. Our implementation can be found at: https://***/Att100/DAANet. © 2023 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prediction and scheduling for blast furnace gas generation based on time series feature extraction

Prediction and scheduling for blast furnace gas generation b...

引用

Industrial Cyber-Physical systems (ICPS)

作者： Huihang Li Jie Hu Qingfeng Yang Luefeng Chen Min Wu School of Automation China University of Geosciences Wuhan China Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Wuhan China Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China

Due to the significant time lag and under-regulation, predicting the blast furnace gas generation and formulating its scheduling strategy is complex. This paper proposes a blast furnace gas generation prediction method based on time series feature extraction and designs a blast furnace gas scheduling strategy based on the prediction results. Firstly, Pearson correlation analysis is used to identify the parameters that have a significant correlation with the blast furnace gas generation, and the selected parameters are decomposed into several intrinsic mode components with different frequency characteristics using the complete ensemble empirical mode decomposition; Then, the principal component analysis method is used to extract the principal components of several intrinsic modal components, and these principal components are employed as the inputs of long short-term memory neural network to predict the blast furnace gas generation; Finally, according to the prediction results designs the scheduling strategy of blast furnace gas. The experiment and contrast experiments are carried out with the industrial field data, and experimental results illustrate that the proposed method is correct and effective.

关键词： Job shop scheduling Correlation Time series analysis Neural networks Predictive models Blast furnaces Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Towards Time-Variant-Aware Link Prediction in Dynamic Graph Through Self-supervised Learning 9th

Towards Time-Variant-Aware Link Prediction in Dynamic Graph...

引用

19th International Conference on Advanced Data Mining and Applications, ADMA 2023

作者： Wen, Guangqi Cao, Peng Jin, Zhiyong Song, Ruoxian Liu, Xiaoli Yang, Jinzhu Zaiane, Osmar R. Computer Science and Engineering Northeastern University Shenyang China Key Laboratory of Intelligent Computing in Medical Image of Ministry of Education Northeastern University Shenyang China National Frontiers Science Center for Industrial Intelligence and Systems Optimization Shenyang110819 China DAMO Academy Alibaba Group Hangzhou China Alberta Machine Intelligence Institute University of Alberta EdmontonAB Canada

ISBN: (纸本)9783031466731

Dynamic graph link prediction is a challenging problem because the graph topology and node attributes vary at different times. A purely supervised learning scheme for the dynamic graph data usually leads to poor generalization due to insufficient supervision. As a promising solution, self-supervised learning can be introduced to dynamic graph analysis tasks. However, the self-supervised learning paradigm for dynamic graph learning has not been sufficiently investigated due to the complicated properties of the evolving graphs. We assume that the dynamic graphs consist of three independent types of key factors, i.e., graph time-variant information, time-invariant information and noise. Based on this assumption, we propose the Self-supervised Decoupling for Dynamic Graph (SDDG) framework for explicitly decoupling the representation which characterizes these three factors, thus enhancing the interpretability of the learned representation and link prediction performance. More specifically, we design an encoder-decoder architecture to sufficiently exploit the dynamic graph itself with multiple regularizations, so that the time-variant embedding of dynamic graph data can be effectively decoupled from the perspectives of node and structure for time-variant-aware link prediction. Experiments on five benchmark link prediction tasks demonstrate the significant improvement of our SDDG over the state-of-the-art methods. For example, SDDG achieves 98.2% and 97.3% Top-1 AUC on Reddit-B and Reddit-T, outperforming the DDGCL model, by 1.9% and 1.8%, respectively. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Supervised learning

来源：评论

学校读者我要写书评

暂无评论

Trend Features Extraction of Apparent Resistivity in Tunnel Engineering

Trend Features Extraction of Apparent Resistivity in Tunnel ...

引用

Chinese Automation Congress (CAC)

作者： Kai Zhou Weihua Cao Chao Gan Yupeng Li School of Automation China University of Geosciences Wuhan China Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Wuhan China Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of Education Wuhan China

Troublesome incidents like sudden water inflows increase the risk of collapse accidents in tunnel excavation. In this study, a data-driven underground water prediction method is proposed based on trend features extracted from apparent resistivity. A novel framework is developed for extracting trend features from the contour lines of apparent resistivity. These trend features are subsequently integrated with numerical features from the resistivity matrix for classification. The effectiveness of the proposed method is demonstrated by apparent resistivity data from real tunnel engineering. The result indicates that the classification accuracy of the proposed method outperforms the method without feature extraction.

关键词：

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：