检索结果-内蒙古大学图书馆

Transformer-based multi-level attention integration network for video saliency prediction

Multimedia Tools and Applications 2025年第13期84卷 11833-11854页

作者： Tan, Rui Sun, Minghui Liang, Yanhua Software College Jilin University Qianjin Street Jilin Changchun130012 China College of Computer Science and Technology Jilin University Qianjin Street Jilin Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Qianjin Street Jilin Changchun130012 China

Most existing models for video saliency prediction heavily rely on 3D convolutional operations to extract spatio-temporal features. However, it is worth noting that 3D convolution produces a local receptive field, which may struggle to capture long-range spatio-temporal dependencies effectively. To compensate for such shortage, this paper introduces a novel approach called the Transformer-based Multi-level Attention Integration Network (TMAI-Net) for video saliency prediction. TMAI-Net is designed as a two-stream encoder-decoder model, carefully integrating multi-level features of semantic information. Our model incorporates a Multi-level Interactive Attention(MLIA) module and a Transformer, both implemented based on self-attention mechanism, which are placed at different levels of the model to capture long-range spatio-temporal feature dependencies. Additionally, our model operates on input video frames and attentional patches, allowing the Transformer module to capture structural similarities between related objects in global features and attention features. This, in turn, enables the model to allocate increased attention to salient areas. The efficacy of our proposed approach is validated through extensive experiments conducted on three widely recognized benchmark datasets. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

A Neural Network Framework for Classifying Educational Content Aligned with Learning Outcomes

引用

Procedia computer science 2025年 257卷 529-536页

作者： Adel Alkhalil Department of Software Engineering College of Computer Science and Engineering University of Ha’il Ha’il 81481 Saudi Arabia

Rapid growth of digital educational content necessitates efficient and accurate methods for organizing and mapping resources to ensure well-alignment with targeted learning outcomes, academic standards, and competency frameworks. Traditional text classification approaches, including rule-based and classical machine learning techniques, often fail to address the semantic diversity and scalability demands of modern educational systems. This study investigates the application of neural networks for text classification to automate the mapping of educational content into predefined categories. Leveraging state-of-the-art architectures such as Long Short-Term Memory (LSTM) networks, and transformers like BERT, we present an architecture of a systematic Classification of educational materials. We discuss the implications of this work for adaptive learning environments, emphasizing the potential of neural networks to enhance the efficiency and scalability of content mapping. This study contributes to the growing body of research in artificial intelligence for education and sets the stage for further exploration into multilingual and domain-specific content classification methods.

关键词： Neural Networks Text Classification Educational Content Curriculum Mapping Transformer Models Learning Outcomes

来源：评论

学校读者我要写书评

暂无评论

MoreGCN: Distributed IoT Service Recommendation Considering Temporal User Interest Dynamics

引用

IEEE Internet of Things Journal 2025年第10期12卷 13657-13670页

作者： Zhou, Yu Hu, Chunqiang Liu, Zewei Xing, Xiaoshuang Li, Xingwang Chongqing University School of Big Data and Software Engineering Chongqing401331 China China Southern Power Grid Joint Laboratory on Cyberspace Security Guangdong510663 China Changshu Institute of Technology School of Computer Science and Engineering Suzhou215556 China Henan Polytechnic University College of Physics and Electronic Information Engineering Jiaozuo454000 China

With the continuous development of Internet of Things (IoT), significant value has been generated, but numerous challenges remain. Recommender systems, as an effective tool to optimize IoT services, can significantly enhance user experience. However, the IoT’s demands for low latency and high-computational load make it difficult for traditional recommender systems to adapt. Moreover, traditional approaches often overlook the dynamic nature of user preferences, which are crucial for determining user satisfaction with IoT services. To address these issues, we propose a novel method called MoreGCN, which quantifies the temporal evolution of user preferences and integrates user interest modeling to accurately match similar users. This approach guides the learning of convolutional networks during the recommendation process. Deployed within a distributed computing framework and combined with meta computing, MoreGCN significantly improves computational efficiency and recommendation accuracy. Experimental results demonstrate that, across three benchmark datasets, MoreGCN consistently outperforms several existing state-of-the-art methods in terms of performance. © 2014 IEEE.

关键词： Internet of Things Recommender systems Computational modeling Accuracy Load modeling Edge computing Feature extraction Collaborative filtering Vectors Servers

来源：评论

学校读者我要写书评

暂无评论

Lift: Exploiting Hybrid Stacked Memory for Energy-Efficient Processing of Graph Convolutional Networks 23

Lift: Exploiting Hybrid Stacked Memory for Energy-Efficient ...

引用

Proceedings of the 60th Annual ACM/IEEE Design Automation Conference

作者： Jiaxian Chen Zhaoyu Zhong Kaoyi Sun Chenlin Ma Rui Mao Yi Wang College of Computer Science and Software Engineering Shenzhen University Shenzhen China

ISBN: (纸本)9798350323481

Graph Convolutional Networks (GCNs) are powerful learning approaches for graph-structured data. GCNs are both computing- and memory-intensive. The emerging 3D-stacked computation-in-memory (CIM) architecture provides a promising solution to process GCNs efficiently. The CIM architecture can provide near-data computing, thereby reducing data movement between computing logic and memory. However, previous works do not fully exploit the CIM architecture in both dataflow and mapping, leading to significant energy *** paper presents Lift, an energy-efficient GCN accelerator based on 3D CIM architecture using software and hardware co-design. At the hardware level, Lift introduces a hybrid architecture to process vertices with different characteristics. Lift adopts near-bank processing units with a push-based dataflow to process vertices with strong re-usability. A dedicated unit is introduced to reduce massive data movement caused by high-degree vertices. At the software level, Lift adopts a hybrid mapping to further exploit data locality and fully utilize the hybrid computing resources. The experimental results show that the proposed scheme can significantly reduce data movement and energy consumption compared with representative schemes.

关键词： 3D-stacked memory

来源：评论

学校读者我要写书评

暂无评论

Dealing with Noisy Data in Federated Learning: An Incentive Mechanism with Flexible Pricing 34

Dealing with Noisy Data in Federated Learning: An Incentive ...

引用

34th ACM Web Conference, WWW 2025

作者： Wang, Hengzhi Chen, Haoran Ma, Minghe Cui, Laizhong Shenzhen University College of Computer Science and Software Engineering Shenzhen China

ISBN: (纸本)9798400712746

Federated Learning (FL) has emerged as a promising training framework that enables a server to effectively train a global model by coordinating multiple devices, i.e., clients, without sharing their raw data. Keeping data locally can ensure data privacy, but also makes the server difficult to assess data quality, leading to the noisy data issue. Specifically, for any given training task, only a portion of each client’s data is relevant and beneficial, while the rest may be redundant or noisy. Training with excessive noisy data can degrade performance. Motivated by this, we investigate the limitations of existing studies and develop an incentive mechanism with flexible pricing tailored for noisy data settings. The insight lies in mitigating the impact of noisy data by selecting appropriate clients and incentivizing them to clean their data spontaneously. Further, both rigorous theoretical analysis and extensive simulations compared with state-of-the-art methods have been well-conducted to validate the effectiveness of the proposed mechanism. © 2025 Copyright held by the owner/author(s).

关键词： Federated learning incentive mechanism noisy data

来源：评论

学校读者我要写书评

暂无评论

Leveraging Chain of Thought for Automated Medical Coding of Lymphoma Cases 10th

Leveraging Chain of Thought for Automated Medical Coding o...

引用

10th China Health Information Processing Conference, CHIP 2024

作者： Song, Wenlong Hu, Jiaxin Li, Zixuan Yuan, Chi College of Computer Science and Software Engineering Hohai University Nanjing China

ISBN: (纸本)9789819642977

Objective: This study aims to explore the application of Chain of Thought (CoT) reasoning in automating ICD coding, specifically focusing on lymphoma cases. By leveraging large language models (LLMs) and CoT prompting, the research aims to improve the accuracy and interpretability of automated ICD coding in healthcare. Methods: We optimized CoT prompting for Qwen 2.5:14b to guide the model in reasoning through the diagnostic process in a step-by-step manner. There are three main components for our pipeline. This framework was then evaluated on multiple LLMs, including Qwen 2.5, Llama 3.1, and Gemma 2, using a clinical lymphoma dataset for automated ICD-10 and ICD-O coding tasks. The models were assessed on their ability to extract relevant diagnoses from clinical text and map them to appropriate ICD codes. Results: The results indicated that CoT reasoning improves the interpretability and accuracy of ICD coding. However, the generalizability of CoT-based prompt design is not guaranteed for all the LLMs. Prompts optimized for one model do not consistently transfer to others, highlighting the importance of customizing prompts for each specific model. Conclusion: Our findings suggest that CoT reasoning can enhance automated ICD coding when tailored to the characteristics of each model. Future work should focus on optimizing CoT prompting techniques for different models to enhance generalizability and improve the scalability of automated coding systems in healthcare. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Oncology

来源：评论

学校读者我要写书评

暂无评论

Research on Chip Special Layout Recognition Technology 24

Research on Chip Special Layout Recognition Technology

引用

Proceedings of the 2024 4th International Conference on Big Data, Artificial Intelligence and Risk Management

作者： Jimao Jia Software Engineering in College of Computer Science College of Software Inner Mongolia University Hohhot Inner Mongolia China

ISBN: (纸本)9798400709753

Chip analysis is an important means of chip research, and with the progress of the chip process, the traditional means are subject to great challenges. For this reason, the research and application of chip automation and intelligent technology is an important topic. In this paper, for the chip analysis in the layout presents special process holes, missing effective elements, poor image quality and other characteristics, the existing recognition algorithms have been applied to the difficulties of bad results, proposed deep learning-based improvement algorithms, basically solves the layout of the large holes, hollow holes, dark holes, and other special elements of the problem, the recognition effect reaches the expected.

关键词： chip analysis

来源：评论

学校读者我要写书评

暂无评论

Adversarial Knowledge Transfer for Black-Box Model Inversion Attack

Adversarial Knowledge Transfer for Black-Box Model Inversion...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Xinhao Liu Zetao Lin Yingzhao Jiang Qiao Yan College of Computer Science & Software Engineering Shenzhen University Shenzhen China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Recent advancements in model inversion attacks have raised privacy concerns, exploiting access to models to reconstruct private training data from inputs and outputs. These attacks are categorized as white-box, black-box, or label-only based on access level. We propose a new black-box model inversion attack, Label-Controlled Adversarial Knowledge Transfer (L-AdKT). L-AdKT leverages adversarial training with a Generative Adversarial Network (GAN) and a substitute model to extract knowledge from the target model. The substitute model minimizes discrepancies with the target model while guiding the generator to produce realistic samples. This approach enables white-box techniques to be applied in black-box settings. Experiments show that L-AdKT outperforms state-of-the-art black-box attacks by over 20% across benchmarks and remains robust against various defense mechanisms.

关键词： Training Closed box Training data Signal processing Generative adversarial networks Data models Speech processing Knowledge transfer Glass box Resilience

来源：评论

学校读者我要写书评

暂无评论

Human Action Recognition in Multi-Level Convolutional Temporal Attention Network

Human Action Recognition in Multi-Level Convolutional Tempor...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Qian Huang ZhongQi Chen Chang Li WeiWen Qian College of Computer Science and Software Engineering Hohai University NanJing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Human Action Recognition (HAR) has widespread applications in areas such as human-computer interaction, elderly care, and home healthcare. However, current sensor-based HAR faces challenges of low fine-grained recognition performance and difficulty in distinguishing similar actions. To solve this problem, this paper proposes a model based on Multilevel Convolutional Time Series Attention Network (MCTSANet). By Multi-ResCNN to pay attention to different levels of features and using Time Series Attention (TSA) to pay attention to the more important data in the channel, so as to improve the ability of confusable action recognition. Experiments on three public datasets show that the proposed method outperforms state-of-the-art sensor-based HAR approaches.

关键词： Human computer interaction Convolution Time series analysis Noise Speech recognition Medical services Human activity recognition Speech processing Older adults Faces

来源：评论

学校读者我要写书评

暂无评论

Spatially Compact Dense Block Mining in Spatial Tensors 25

Spatially Compact Dense Block Mining in Spatial Tensors

引用

Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1

作者： Weike Tang Dingming Wu Tsz Nam Chan Kezhong Lu College of Computer Science & Software Engineering Shenzhen University Shenzhen China

ISBN: (纸本)9798400712456

Spatial tensors have been extensively used in a wide range of applications, including remote sensing, geospatial information systems, conservation planning, and urban planning. We study the problem of Spatially Compact Dense (SCD) block mining in a spatial tensor, which targets for discovering dense blocks that cover small spatial regions. However, most of existing dense block mining (DBM) algorithms cannot solve the SCD-block mining problem since they only focus on maximizing the density of candidate blocks, so that the discovered blocks are spatially loose, i.e., covering large spatial regions. Therefore, we first formulate the problem of mining top-k Spatially Compact Dense blocks (SCD-blocks) in spatial tensors, which ranks SCD-blocks based on a new scoring function that takes both the density value and the spatial coverage into account. Then, we adopt a filter-refinement framework that first generates candidate SCD-blocks with good scores in the filtering phase and then uses the traditional DBM algorithm to further maximize the density values of the candidates in the refinement phase. Due to the NP-hardness of the problem, we develop two types of solutions in the filtering phase, namely the top-down solution and the bottom-up solution, which can find good candidate SCD-blocks by approximately solving the new scoring function. The evaluations on four real datasets verify that compared with the dense blocks returned by existing DBM algorithms, the proposed solutions are able to find SCD-blocks with comparable density values and significantly smaller spatial coverage.

关键词： dense block

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：