检索结果-内蒙古大学图书馆

Proceedings of the 4th International Conference on Neural Computing for Advanced Applications, NCAA 2023

作者： Lu, Yao Yin, Xiaoxu Qin, Feng Huang, Ke Zhang, Menghua Huang, Weijie School of Electrical Engineering University of Jinan Jinan250000 China Qilu Aerospace Information Research Institute Jinan China Zaozhuang Vocational College of science and technology Zaozhuang China School of Information Science and Engineering Shandong Normal University Jinan250000 China

ISBN: (纸本)9789819958467

In recent years, the performance of visual inertial odometry (VIO) based on deep learning has shown significant advantages over traditional geometric methods. However, all existing methods estimate each pose through visual and inertial measurements, which involves a large amount of computational redundancy, resulting in huge time costs and hardware damage when training and deploying on devices. In order to maintain accuracy while reducing the number of training parameters, an improved algorithm based on Visual-selective-VIO is proposed. To reduce the number of network parameters and maintain the training accuracy, a unique attention mechanism is designed for the visual branch and a lightweight pose estimation module. By improving the visual branch, we serialize the information of attention feature maps, covering both channel and spatial dimensions. Then, we multiply these two feature maps with the original input feature maps for adaptive feature correction. This method improves the sensitivity of the model to channel features and enables more accurate image localization. Experimental results show that our algorithm maintains accuracy with a 10 % reduction in network parameters compared to advanced VIO algorithm, making it more suitable for training large-scale datasets and deployment in practical applications. © 2023, The Author(s), under exclusive license to springer Nature singapore Pte Ltd.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

ssCCPC-Net: simultaneously learning 2D and 3D Features with CLIP for semantic scene Completion on Point Cloud 41st

SSCCPC-Net: Simultaneously Learning 2D and 3D Features with...

引用

41st computer Graphics International Conference, CGI 2024

作者： Duan, Wantong Bao, Yongtang Qi, Yue State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and Engineering Beihang University Beijing100191 China Shandong University of Science and Technology Qingdao266590 China Jingdezhen Research Institute of Beihang University Jingdezhen333000 China Qingdao Research Institute of Beihang University Qingdao266000 China

ISBN: (纸本)9783031820236

Compare to traditional scene Completion (sC), semantic scene Completion (ssC) is a challenging task that aims to generate complete and semantically consistent 3D scene from partial and sparse input data, which is fundamental to fully understanding the scene and being able to interact with it. Consequently, the ssC task has received much attention in recent years. Most of the methods are voxel-based approaches, but they have high computational and memory requirements. A few works based on point cloud do not sufficiently exploit the correlation between semantic segmentation and geometric completion subtasks, while focusing too much on point cloud shape features and ignoring the rich texture information that RGB images can provide. In this paper, we present ssCCPC-Net (semantic scene Completion with CLIP on Point Cloud-Net), a novel network architecture for point cloud semantic scene completion using a combination of 2D and 3D features. Inspired by recent works of large pretrained vision-language models in semantic segmentation, we explore to accomplish ssC task with the help of Contrastive Language-Image Pre-Training (CLIP) model. specifically, we use the CLIP features for guidance to fuse the 2D features extracted from the RGB image and the 3D features extracted from the point cloud. The fused features are then fed into our designed semantic-Completion Decoder for per-point semantic prediction and semantic labeling-assisted point cloud completion. Finally, we obtain the complete semantically point cloud. Numerous experiments have demonstrated that our method has higher effectiveness and generalizability compared to state-of-the-art methods. © The Author(s), under exclusive license to springer Nature switzerland AG 2025.

关键词： semantic segmentation

来源：评论

学校读者我要写书评

暂无评论

Polynomial Adaptation of Large-scale CNNs for Homomorphic Encryption-based secure Inference 8th

Polynomial Adaptation of Large-Scale CNNs for Homomorphic ...

引用

8th International symposium on Cyber security, Cryptology, and Machine learning, CsCML 2024

作者： Baruch, Moran Drucker, Nir Ezov, Gilad Goldberg, Yoav Kushnir, Eyal Lerner, Jenny soceanu, Omri Zimerman, Itamar IBM Research Haifa Israel Bar-Ilan University Ramat Gan Israel AI2 Tel Aviv-Yafo Israel

ISBN: (纸本)9783031769337

Enabling secure inference of large-scale CNNs using Homomorphic Encryption (HE) requires a preliminary step for adapting unencrypted pre-trained models to only use polynomial operations. Prior art advocates for high-degree polynomials for accurate approximations, which comes at the price of extensive computations. We demonstrate that low-degree polynomials can be sufficient for accurate approximation even for large-scale DNNs. For that, we introduce a dedicated fine-tuning process on unencrypted data that reduces the input range to the activation functions. The resulting models have competitive accuracy of up to 3.5% degradation from the original non-polynomial model, which outperforms prior art on tasks such as ImageNet classification over ResNet and ConvNeXt. Upon adaptation, these models can process HE-encrypted samples and are ready for secure inference. based on these, we provide optimization insights for activation functions and skip connections, enhancing HE evaluation efficiency. We evaluated ResNet50-152 on encrypted ImageNet samples, an accomplishment not previously reached by polynomial networks, in just 3:13–7:12 min, using commodity hardware under the CKKs scheme with 128-bit security. In comparison to prior high-degree polynomial solutions, our low-degree polynomials boost evaluation latency, for example, by 3× for ResNet-50 and CIFAR-10. We further show our approach versatility, by adapting the CLIP model for secure zero-shot predictions, highlighting new potential in HE and transfer learning. © The Author(s), under exclusive license to springer Nature switzerland AG 2025.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Mental state Recognition Through speech Analysis Using Bidirectional Long short-Term Memory Network and Convolutional Neural Network Hybrid Model 10

Mental State Recognition Through Speech Analysis Using Bidir...

引用

10th International Conference on smart Computing and Communication, ICsCC 2024

作者： Erquiza, Axl Julian B. Mamaclay, Russel Janzen E. Platon, Elisha Francia L. Acula, Donata D. University of Santo Tomas Department of Computer Science Manila Philippines Research Center for Natural and Applied Sciences University of Santo Tomas Department of Computer Science Manila Philippines

ISBN: (纸本)9798350363104

This research investigates the effectiveness of Convolutional Neural Networks (CNNs) and Bidirectional Long short-Term Memory (Bi-LsTM) networks in recognizing mental states through speech analysis. Leveraging the DAIC-WoZ audio dataset, the researchers classified mental states across five levels and examined the performance of three CNN architectures: ResNet50, VGGNet16, and InceptionV3, as well as their Bi-LsTM hybrids. The findings indicate varying degrees of effectiveness when integrating Bi-LsTM layers: ResNet50 with Bi-LsTM improved in precision but not in training accuracy, VGGNet16 maintained its performance levels, while InceptionV3 showed negligible benefits from Bi-LsTM integration. The study's outcomes, derived from statistical analyses and performance evaluations, suggest that Bi-LsTM layers do not consistently enhance the predictive accuracy of CNN s in speech-based mental state analysis. This highlights the complexity of choosing and optimizing deep learning architectures for specific tasks in mental health assessment. The research contributes to this field by outlining the potentials and limitations of advanced neural network methodologies in speech analysis for mental health assessment, emphasizing the nuanced role of model architecture in such applications. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Effects of selection of the learning set Formation strategy and Filtration Method on the Effectiveness of a BCI based on Near Infrared spectrometry

引用

Neuroscience and Behavioral Physiology 2023年第3期53卷 373-380页

作者： Isaev, M.R. Bobrov, P.D. Institute of Higher Nervous Activity and Neurophysiology Russian Academy of Sciences Moscow Russian Federation Pirogov Russian National Research Medical University Russian Ministry of Health Moscow Russian Federation

We propose an algorithm for the operation of a brain–computer interface based on recording hemodynamic activity using near-infrared spectrometry (NIRs) adapted for use in the rehabilitation of motor disorders. The algorithm includes a filtration method that takes account of the frequency of presentation of instructions with the aim of minimizing the time delay in the target frequency range, sequential classification of the states of rest and the mental tasks being performed, as well as additional training of the interface classifier using previously obtained data. The influences of the proposed methods for filtration, additional training, and selection of a smaller number of channels on the accuracy of recognition of motor imagery is evaluated using data from three series of experiments previously conducted in the laboratory with healthy volunteers. Removal of low-frequency noise components from the NIRs signal using the proposed filtration method is shown to produce significant increases in classification accuracy, as did use of additional training from previous sessions in the same person;the number of recording channels could be decreased without loss of recognition accuracy. © 2023, springer Nature switzerland AG.

关键词： brain–computer interface filtration motor imagery near-infrared spectrometry

来源：评论

学校读者我要写书评

暂无评论

research on Improving the Detection Method of Welding Defects in Drum Welding of CoT-YOLOv7 4

Research on Improving the Detection Method of Welding Defect...

引用

4th International Conference on computer Vision and Data Mining, ICCVDM 2023

作者： Hu, Wenyuan Zhai, Yusheng Yun, Ruiguang Huang, Qiulai China Coal Research Institute Beijing100013 China Tiandi Shanghai Mining Equipment Technology Co. Ltd. Shanghai201400 China

ISBN: (纸本)9781510674448

In response to the problem of low efficiency and accuracy in detecting welding defects on the surface of coal mining machine drums, a drum welding defect automatic detection method based on the improved YOLOv7 algorithm is proposed. By utilizing the characteristic of long-distance fusion of multiple feature information in the Transformer model, the global context CoT model is introduced into the YOLOv7 algorithm network, and experimental verification is conducted. The results show that this detection method has high detection accuracy in welding defect detection and can eliminate the image redundancy problem caused by the introduction of attention mechanism. It can be applied to the quality inspection of coal mining machine drum welding products. © 2024 sPIE.

关键词： Coal mines

来源：评论

学校读者我要写书评

暂无评论

Natural Language Processing for the Turkish Academic Texts in the Engineering Field: Key-Term Extraction, similarity Detection, subject/Topic Assignment 19th

Natural Language Processing for the Turkish Academic Texts i...

引用

19th International Conference on Artificial Intelligence Applications and Innovations (AIAI)

作者： Kat, Bora Middle East Tech Univ Ankara Turkiye Sci & Technol Res Council Turkey TUBITAK Ankara Turkiye

ISBN: (纸本)9783031341090;9783031341076;9783031341069

The information retrieved from texts plays crucial roles in many aspects. Although there are significant attempts on natural language processing for various types of texts in Turkish, none of them deals with academic texts. This study mainly aims to retrieve precise key terms from Turkish academic texts in the field of engineering and develops algorithms for similarity detection and automatic classification based on these key terms. In the first step of this study: a library and customized templates, that can transform the n-grams into structured forms, are created by considering the features of engineering terminology and the grammar of the Turkish language. Then, a customized similarity detection algorithm is developed. Finally, the Naive Bayes Classifier is used to assign the documents to the appropriate engineering sub-fields. The project proposals submitted to The scientific and Technological research council of Turkey (***) Academic research Funding Program Directorate (ARDEB) are analyzed as a case study. The results indicate that the proposed similarity algorithm correctly detects almost all of the re-submitted proposals while the accuracy of the classifier is 83.3% in the first prediction and reaches up to 96.4% in the first three predictions over a sample of 1255 proposals.

关键词： Key term extraction Feature extraction Natural language processing (NLP) supervised machine learning Naive Bayes classifier TUB.ITAK Conceptual similarity subject/topic assignment

来源：评论

学校读者我要写书评

暂无评论

CCLNet: Causal and Contrastive learning Framework for Enhanced Pulmonary Embolism Detection

CCLNet: Causal and Contrastive Learning Framework for Enhanc...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Wang, Zhipeng Ge, Ruiquan Yu, Jianxun Qin, Feiwei Tian, Yuan Li, Nannan Min, Wenwen Elazab, Ahmed Wang, Changmiao Hangzhou Dianzi University HDU-ITMO Joint Institute Hangzhou China Hangzhou Dianzi University School of Computer Science Hangzhou China Dalian Polytechnic University School of Management Dalian China Chinese University of Hong Kong Second Affiliated Hospital Shenzhen China Macau University School of Science and Technology China Yunnan University School of Information Science and Engineering Yunnan China Shenzhen University School of Biomedical Engineering Shenzhen China Shenzhen Research Institute of Big Data Medical Big Data Lab Shenzhen China

ISBN: (纸本)9798350386226

The fusion of multimodal medical data is crucial for helping doctors make accurate treatment decisions. For example, combining Computed Tomography Pulmonary Angiography (CTPA) with Electronic Health Records (EHR) can significantly improve the accuracy of Pulmonary Embolism (PE) detection, thereby increasing patient survival rates. Although multimodal learning has advantages in PE diagnosis, the heterogeneity of multimodal data poses a significant challenge to accurate diagnosis. The natural semantic and structural differences between data modalities make it difficult to effectively integrate their information. In addition, within a single modality, the existence of redundant and irrelevant information introduces unnecessary variability, making the data more complex, and making stable diagnosis challenging. To address these issues, we propose a new framework called CCLNet, which includes a contrastive learning component for addressing inter-modality heterogeneity and a causal learning component for handling intra-modality heterogeneity. specifically, we achieve precise alignment between visual and tabular modalities by using global-level information to soften labels during contrastive learning. In addition, by using causal intervention methods to eliminate the influence of heterogeneous factors within the modality, we can accurately reveal the causal relationship between features and targets, thereby improving the accuracy and stability of the model. Experimental results demonstrate that our method performs excellently, achieving the best results. Our code is available at https://***/LeavingstarW/CLPE. © 2024 IEEE.

关键词： Contrastive learning

来源：评论

学校读者我要写书评

暂无评论

A Pan-European Review of Good Practices in Early Intervention safeguarding Practice with Children, Young People and Families: Evidence Gathering to Inform a Multi-disciplinary Training Programme (the ERICA project) in Preventing Child Abuse and Neglect in seven European Countries

引用

International Journal on Child Maltreatment: research, Policy and Practice 2023年第1期6卷 119-130页

作者： Appleton, J.V. Bekaert, s. Hucker, J. Zlatkute, G. Paavilainen, E. schecke, H. specka, M. scherbaum, N. Jouet, E. Zabłocka-Żytka, L. Woźniak-Prus, M. Czabała, J. Cz. Kluczyńska, s. Bachi, B. Bartoli, F. Carrà, G. Cioni, R.M. Crocamo, C. Rantanen, H.E. Kaunonen, M. Nieminen, I. Roe, L. Keenan, K. Viganò, G. Baldacchino, A. Oxford United Kingdom Oxford School of Nursing and Midwifery Faculty of Health and Life Sciences Oxford Brookes University Oxford United Kingdom School of Medicine University of St Andrews St Andrews United Kingdom Faculty of Social Sciences/Health Sciences Unit Etelä-Pohjanmaa Hospital District Tampere University Tampere Finland LVR-Hospital Essen Department of Addictive Behaviour and Addiction Medicine Medical Faculty University of Duisburg-Essen Essen Germany Laboratoire de Recherche en Santé Mentale Et Sciences Humaines Et Sociales Groupe Hospitalier Universitaire Paris Psychiatrie & Neurosciences (GHU- PARIS) Paris France Department of Psychology The Maria Grzegorzewska University Warsaw Poland Department of Psychology University of Warsaw Warsaw Poland Department of Medicine and Surgery University of Milano-Bicocca Milan Italy School of Geography and Sustainable Development University of St Andrews St Andrews United Kingdom Synergia S.R.L. and Department of Decision Sciences Università L. Bocconi Milan Italy Formerly Oxford Institute of Nursing Midwifery and Allied Health Research (OxINMAHR) Faculty of Health and Life Sciences Oxford Brookes University Oxford United Kingdom Faculty of Social Sciences/Health Sciences Unit Pirkanmaa Hospital District Tampere University Tampere Finland

Child maltreatment has detrimental social and health effects for individuals, families and communities. The ERICA project is a pan-European training programme that equips non-specialist threshold practitioners with knowledge and skills to prevent and detect child maltreatment. This paper describes and presents the findings of a rapid review of good practice examples across seven participating countries including local services, programmes and risk assessment tools used in the detection and prevention of child maltreatment in the family. learning was applied to the development of the generic training project. A template for mapping the good practice examples was collaboratively developed by the seven participating partner countries. A descriptive data analysis was undertaken organised by an a priori analysis framework. Examples were organised into three areas: programmes tackling child abuse and neglect, local practices in assessment and referral, risk assessment tools. Key findings were identified using a thematic approach. seventy-two good practice examples were identified and categorised according to area, subcategory and number. A typology was developed as follows: legislative frameworks, child health promotion programmes, national guidance on child maltreatment, local practice guidance, risk assessment tools, local support services, early intervention programmes, telephone or internet-based support services, COVID-19 related good practices. Improved integration of guidance into practice and professional training in child development were highlighted as overarching needs. The impact of COVID-19 on safeguarding issues was apparent. The ERICA training programme formally responded to the learning identified in this international good practice review. © 2022, The Author(s).

关键词： Child maltreatment Child protection European Multidisciplinary safeguarding Training

来源：评论

学校读者我要写书评

暂无评论

Design and implementation of small infrastructure digital review system based on BIM 4

Design and implementation of small infrastructure digital re...

引用

4th International Conference on computer Vision and Pattern Analysis, ICCPA 2024

作者： Zhao, Chunhui Wang, Yue Zhang, Jiyong Zhang, Zhuoqun Li, Danli Liu, Donghui State Grid Economic and Technological Research Institute Co. Ltd. Beijing China Fujian Great Power Science and Technology Co. Ltd. Beijing China

ISBN: (纸本)9781510682528

In response to the characteristics of a large number of small infrastructure project review, wide professional involvement, and strong review timeliness, this paper proposes a design scheme for a digital review system for small infrastructure projects based on BIM (Building Information Modeling) technology. It focuses on the five aspects of "human, material, machine, method, and environment" in small infrastructure projects to achieve intelligent three-dimensional review. based on a rule engine, it realizes one click extraction of key indicators for review materials, vigorously promotes standardization and standardization of review, enhances the authority and effectiveness of review, effectively accumulates and forms digital assets for review, and realizes the efficient transformation from manual review to human-machine collaboration. © 2024 sPIE.

关键词： Building Information Model

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：