检索结果-内蒙古大学图书馆

20th International computer conference on Wavelet Active Media Technology and Information Processing, ICCWAMTIP 2023

作者： Jiayang, Gu Fengming, Yao The School of Computer Science and Engineering University of Electronic Science and Technology of China Chengdu611731 China The School of Aeronautics and Astronautics University of Electronic Science and Technology of China Chengdu611731 China

ISBN: (纸本)9798350318982

Video captioning models easily suffer from long-tail distribution of phrases, which makes captioning models prone to generate vague sentences instead of accurate ones. However, existing debiasing strategies tend to export external knowledge to build dependency trees of words or refine frequency distribution by complex losses and extra input features, which lack interpretability and are hard to train. To mitigate the impact of granularity bias on the model, we introduced a statistical-based bias extractor. This extractor quantifies the information content within sentences and videos, providing an estimate of the likelihood that a video-sentence pair is affected by granularity bias. Furthermore, with the growing trend of integrating contrastive learning methods into video captioning tasks, we use a bidirectional triplet loss to get more negative samples in a batch. Subsequently, we incorporate the margin score into the contrastive learning loss, establishing distinct training objectives for head and tail sentences. This approach facilitates the model's training effectiveness on tail samples. Our simple yet effective loss, incorporating Granularity bias, is referred to as the Margin-Contrastive Loss (GMC Loss). The proposed model demonstrates state-of-the-art performance on MSRVTT with a CIDEr of 57.17, and MSVD, where CIDEr reaches up to 138.68. © 2023 IEEE.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

LAME:Layout-Aware Metadata Extraction Approach for Research Articles

引用

computers, Materials & Continua 2022年第8期72卷 4019-4037页

作者： Jongyun Choi Hyesoo Kong Hwamook Yoon Heungseon Oh Yuchul Jung Department of Computer Engineering Kumoh National Institute of Technology(KIT)GumiKorea Korea Institute of Science and Technology Information(KISTI) DaejeonKorea School of Computer Science and Engineering Korea University of Technology and Education(KOREATECH)CheonanKorea

The volume of academic literature,such as academic conference papers and journals,has increased rapidly worldwide,and research on metadata extraction is ***,high-performing metadata extraction is still challenging due to diverse layout formats according to journal *** accommodate the diversity of the layouts of academic journals,we propose a novel LAyout-aware Metadata Extraction(LAME)framework equipped with the three characteristics(e.g.,design of automatic layout analysis,construction of a large meta-data training set,and implementation of metadata extractor).In the framework,we designed an automatic layout analysis using PDF *** on the layout analysis,a large volume of metadata-separated training data,including the title,abstract,author name,author affiliated organization,and keywords,were automatically ***,we constructed a pre-trainedmodel,Layout-Meta BERT,to extract the metadata from academic journals with varying layout *** experimental results with our metadata extractor exhibited robust performance(Macro-F1,93.27%)in metadata extraction for unseen journals with different layout formats.

关键词： Automatic layout analysis layout-MetaBERT metadata extraction research article

来源：评论

学校读者我要写书评

暂无评论

Explainable Network Anomaly Detection with GraphSAGE and SHAP

Explainable Network Anomaly Detection with GraphSAGE and SHA...

引用

7th International conference on Artificial Intelligence in Information and Communication, ICAIIC 2025

作者： Lee, Jihoon Oh, Seungmin Song, Jaeho Noh, Juhyeon Hanh, Minsoo Kim, Jinsul Chonnam National University Dept. of Intelligent Electronics and Computer Engineering GwangJu Korea Republic of

ISBN: (纸本)9798331507022

Network anomaly detection is a critical task for maintaining network stability and security. However, existing models often focus solely on achieving high predictive performance, falling short in providing the interpretability and reliability needed for practical applications. To address this limitation, this study proposes a novel approach that combines GraphSAGE and SHAP. GraphSAGE is designed to classify various types of network anomalies effectively by leveraging network data, while SHAP extracts and quantifies the contributions of key features influencing the model's predictions. The experimental results demonstrate that the proposed model achieved high accuracy and Fl-score, successfully identifying the most significant features for each anomaly class. This study highlights that the integration of GraphSAGE and SHAP enhances the interpretability and practicality of network anomaly detection. By providing clear explanations for the model's predictions, this approach offers actionable insights for network administrators, making it a valuable tool for real-world network management and security applications. © 2025 IEEE.

关键词： Network theory (graphs)

来源：评论

学校读者我要写书评

暂无评论

Learning from Machine Learning and Teaching with Machine Teaching: Using Lessons from Data Science to Enhance Collegiate Classrooms

Learning from Machine Learning and Teaching with Machine Tea...

引用

2023 ASEE Annual conference and Exposition - The Harbor of Engineering: Education for 130 Years, ASEE 2023

作者： Buccafusca, Lucas Department of Electrical and Computer Engineering Johns Hopkins University United States

In the field of data science, advancements in the field of machine learning have led to programs developing high-level reasoning, intricate data understanding, and groundbreaking predictive models. Machine Learning (ML) research aims at making a program 'learn,' that is, develop models and techniques with known information to be able to handle future problems. Traditionally, this is done by increasing the quantity and quality of input data and training the learner in more effective ways to interpret that information. This has a direct parallel to the collegiate classroom, as instructors aim to inspire mastery over a topic to their students through a variety of methods (homework problems, examinations, projects, etc.) and teach them the corresponding skillsets from feedback on these assignments. Machine Teaching (MT) research, on the other hand, aims at making the teacher more productive by using their own cognitive models to improve the quality of the data holistically. Again, this has a corresponding counterpart to current teaching pedagogies;the instructor decides on the details of an assignment from their own knowledge and experience with the end goal of having students retain the information and apply it to future problems. This paper identifies how the various innovations, lessons, and conclusions discovered in the field of artificial intelligence can enhance the quality of a collegiate classroom experience and improve student performance. © American Society for Engineering Education, 2023.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

A Novel Approach for Virtual Locomotion Gesture Classification: Self-Teaching Vision Transformer for a Carpet-Type Tactile Sensor 31

A Novel Approach for Virtual Locomotion Gesture Classificati...

引用

IEEE conference on Virtual Reality and 3D User Interfaces (VR)

作者： Lee, Sung-Ha Joo, Ho-Taek Chung, Insik Park, Donghyeok Choi, Yunho Kim, Kyung-Joong Gwangju Inst Sci & Technol Gwangju South Korea

ISBN: (纸本)9798350374025;9798350374032

Locomotion gesture classification in virtual reality (VR) is the process of analyzing and identifying specific user movements in the real world to navigate virtual environments. However, existing methods often necessitate the use of wearable sensors, which present limitations. To address this, we utilize a high-resolution carpet-type tactile sensor as a foot action recognition interface, which was previously unexplored in the context of locomotion gesture classification. This interface can capture the user's foot pressure data in detail to distinguish similar actions. In this paper, to efficiently process captured user's foot tactile data and classify nuanced actions, we utilize a Vision Transformer (ViT) architecture and propose a novel Self-Teaching Vision Transformer (STViT) model integrating elements of the Shifted window Vision Transformer (SwinViT) and Data-efficient image Transformer (DeiT). However, unlike DeiT, our model uses itself from N-steps prior as the teacher model, which is continuously updated. Therefore, improving the ability to classify actions by referencing its own knowledge from previous training stages progressively refines its understanding of similar action gestures. Also, we used the base architecture of SwinViT to utilize patch merging, which improves the ability to differentiate between variations in similar actions by capturing information at different scales. We evaluated seven vision-based methods, demonstrating promising results. Not only did our model outperform ResNet by 19.6%, but it also outperformed each Deit and SwinViT by 3.3% and 2.9%, achieving 92.7% accuracy. To validate our model's real-world applicability, we conducted user preference tests and in-game performance evaluations with 18 participants. As a result, the participants preferred our model to SwinViT and DeiT, backing up the computational results. The video demonstrating the VR testing for STViT can be found in https://***/NJslvanRn18

关键词： Human-centered computing-Human-computer interaction (HCI)-Interaction techniques-Gestural input Human-centered computing-Human-computer interaction (HCI)-Interaction paradigms-Virtual reality

来源：评论

学校读者我要写书评

暂无评论

An Error-Bounded Space-Efficient Hybrid Learned Index with high Lookup performance 33rd

An Error-Bounded Space-Efficient Hybrid Learned Index with H...

引用

33rd International conference on Database and Expert Systems Applications (DEXA)

作者： Ding, Yuquan Zhao, Xujian Jin, Peiquan Southwest Univ Sci & Technol Mianyang Sichuan Peoples R China Univ Sci & Technol China Hefei Peoples R China

ISBN: (纸本)9783031124266;9783031124259

Learned index is a novel index structure that enhances traditional index structures like B+-tree with machine learning models. It uses a trained model to locate a key position directly without the costly rootto-leaf traversing in the conventional B+-tree. However, the performance of existing learned indices relies on the prediction accuracy of the model. Recently, to address this problem, some existing learned indices, such as PGM-index, set a threshold to ensure that the prediction error is below the threshold. However, the PGM-index introduced additional space costs because it used a bottom-up strategy to build a tree structure to maintain an error threshold at each level. In this paper, we aim to find out a better trade-off between prediction accuracy and space consumption of learned indices. In particular, we propose a new learned index called HLI (Hybrid Learned Index) that offers high prediction accuracy and searching performance with limited space costs. HLI constructs error-bounded leaf nodes to provide high prediction accuracy and searching performance. Meanwhile, it organizes inner nodes without error bounds to reduce extra space costs. Consequently, we demonstrate that HLI can deliver high searching performance with limited space overhead. We conduct experiments on various datasets to compare HLI with several existing indices, including PGM-index, RMI, and B+-tree. The results show that HLI outperforms PGM-index in lookup performance on all datasets by 1.5x on average. In addition, compared to PGM-index, HLI reduces the index size by up to 13x without decaying the lookup performance.

关键词： Learned index Hybrid index Data distribution Piecewise linear approximation

来源：评论

学校读者我要写书评

暂无评论

Classification of Alzheimer's Disease Based on Multi-view Lightweight Neural Network 3

Classification of Alzheimer's Disease Based on Multi-view Li...

引用

3rd International conference on Consumer Electronics and computer Engineering, ICCECE 2023

作者： Zhang, Qiongmin Long, Ying Cai, Hongshun School of Computer Science and Engineering Chongqing University of Technology Chongqing China

ISBN: (纸本)9798350331578

As a degenerative neurodegenerative disease, early detection and intervention are still the key to prevent the aggravation of Alzheimer's disease (AD). computer aided diagnosis based on Magnetic Resonance Imaging (MRI) can provide assistance for early detection of AD. In the application of limited MRI data, a multi-view lightweight neural network is proposed to effectively identify AD by reducing the amount of model computation and parameters and making full use of the image information. In the proposed method, we design an optimized lightweight neural network for feature extraction on limited MRI data. Attention mechanism is adopted to fuse multi-slice features and ensure the consistency of results. For full use of MRI data, personalized weights are learned and applied to fuse the multi-view features of transverse plane, sagittal plane and coronal plane. The experimental results indicate that compare with state-of-The-Art lightweight methods, the proposed multi-view method achieves good balance on parameters (3.75M) and FLOPs (4.45G). Meanwhile, the classification performances of the proposed multi-view method are better in the three AD classification tasks. Through the construction of the model, it is expected to provide a feasible reference lightweight network for further AD detection. © 2023 IEEE.

关键词： Magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

OCCNET: Improving Imbalanced Multi-Centred Ovarian Cancer Subtype Classification in Whole Slide Images 20

OCCNET: Improving Imbalanced Multi-Centred Ovarian Cancer Su...

引用

20th International computer conference on Wavelet Active Media Technology and Information Processing, ICCWAMTIP 2023

作者： Ahmed, Awais Xiaoyang, Zeng Tunio, Muhammad Hanif Butt, Muhammad Hassaan Shah, Syed Attique Chengxiao, Yan Pirzado, Farman Ali Aziz, Abdul School of Computer Science and Engineering University of Electronic Science and Technology of China - UESTC Sichuan611731 China Scchool of Computing and Digital Technology Birmingham City University BirminghamB4 7RQ United Kingdom University Technologico de Monterrey Mexico 64700 Mexico Universidad de Zaragoza Zaragoza50018 Spain

ISBN: (纸本)9798350318982

Ovarian carcinoma is known for its diverse subtypes with unique morphologies and clinical characteristics, causing considerable diagnostic complexities. While deep learning has emerged as a promising avenue for subtype identification, the intrinsic imbalance in different datasets from multiple medical institutions is still a challenging issue. In this paper, we introduce a novel Ensemble Attention Mechanism (EAM) specifically tailored to enhance subtype identification within imbalanced, multicentered ovarian cancer Whole Slide Images (WSIs), namely OCCNet. The OCCNet framework incorporate ensemble learning principles and attention mechanisms to adaptively balance the feature representation and effectively integrate every representation which mitigates inherent imbalances of class distribution. Then, one publicly available extensive Multi-centered Dataset from the ongoing Kaggle Competition is used to evaluate OCCNet's performance. Results show that the proposed method achieved a 93.67% F1-Score and a 93% Balanced Accuracy, revealing its feasibility and superiority. © 2023 IEEE.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

PIOD: An Efficient Parallel Iterative Algorithm for Solving Over-determined Equations 29

PIOD: An Efficient Parallel Iterative Algorithm for Solving ...

引用

29th IEEE International conference on Parallel and Distributed Systems, ICPADS 2023

作者： Tian, Min Wei, Zhenguo Li, Yanlong Du, Wei Xiao, Lei Xu, Chaoshuai China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Computer Networks China Jinan Institute of Supercomputing Technology Jinan Key Laboratory of High Performance Computing China

ISBN: (纸本)9798350330717

The solution of over-determined equations plays a very important role in fields such as data fitting, signal processing, and machine learning. It is of great significance in predicting natural phenomena, optimizing engineering design, and other fields. However, there is currently no efficient method to solve over-determined equations, either being not accurate enough or consuming a lot of time. In this article, we propose a parallel iterative method for solving over-determined equations, called the PIOD algorithm. By using a sub-convergence condition to terminate the iterative calculation, we have developed a task partitioning strategy for the algorithm and implemented parallelization of the solution of over-determined equations on a distributed memory system. Our proposed algorithm achieves an average speedup of 152 ×times compared to the open-source Eigen solver. Additionally, it also achieves a parallel efficiency of over 30%. © 2023 IEEE.

关键词： conjugate gradient method MPI over-determined equations parallel algorithms parallel iterative methods

来源：评论

学校读者我要写书评

暂无评论

Cross-lingual Back-Parsing: Utterance Synthesis from Meaning Representation for Zero-Resource Semantic Parsing

Cross-lingual Back-Parsing: Utterance Synthesis from Meaning...

引用

2024 conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Kang, Deokhyung Hwang, Seonjeong Kim, Yunsu Lee, Gary Geunbae Graduate School of Artificial Intelligence POSTECH Korea Republic of Department of Computer Science and Engineering POSTECH Korea Republic of AiXplain Inc. Los GatosCA United States

ISBN: (纸本)9798891761643

Recent efforts have aimed to utilize multilingual pretrained language models (mPLMs) to extend semantic parsing (SP) across multiple languages without requiring extensive annotations. However, achieving zero-shot cross-lingual transfer for SP remains challenging, leading to a performance gap between source and target languages. In this study, we propose Cross-lingual Back-Parsing (CBP), a novel data augmentation methodology designed to enhance cross-lingual transfer for SP. Leveraging the representation geometry of the mPLMs, CBP synthesizes target language utterances from source meaning representations. Our methodology effectively performs cross-lingual data augmentation in challenging zero-resource settings, by utilizing only labeled data in the source language and monolingual corpora. Extensive experiments on two cross-lingual SP benchmarks (Mschema2QA and Xspider) demonstrate that CBP brings substantial gains in the target language. Further analysis of the synthesized utterances shows that our method successfully generates target language utterances with high slot value alignment rates while preserving semantic integrity. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

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：