检索结果-内蒙古大学图书馆

4th International Conference on Artificial Intelligence, Robotics, and Communication, ICAIRC 2024

作者： Du, Jiayi Zhao, Taiyin Li, Xucheng University of Electronic Science and Technology of China School of Information and Software Engineering Chengdu China

ISBN: (纸本)9798331531225

Federated learning (FL) is widely applied in edge computing scenarios to protect data privacy, but the assumption of independent and identically distributed (IID) data across clients often does not hold in practice, leading to performance degradation. Among the many methods proposed to address data heterogeneity, clustered federated learning (CFL) has emerged as a popular approach. Current CFL methods commonly adopt the expectation-maximization (EM) algorithm, which mitigates the problem of data heterogeneity. However, they still face challenges such as requiring more clients than traditional federated learning and introducing randomness in the clustering process. To overcome these limitations, we propose FedDPC, which integrates the density peak clustering (DPC) algorithm. This approach provides the advantages of not requiring additional clients and having a non-random partitioning strategy. Finally, we conducted experiments to demonstrate the feasibility and effectiveness of our proposed method. © 2024 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Improving the Inference Efficiency of Transformer Models in Machine Translation Tasks by Low-rank Decomposition Methods

Improving the Inference Efficiency of Transformer Models in ...

引用

2024 International Conference on Electronics and Devices, Computational Science, ICEDCS 2024

作者： Liu, Zhijia University of Electronic Science and Technology of China School of Information and Software Engineering Chengdu China

ISBN: (纸本)9798331527624

In recent years, Transformer model has made remarkable progress in machine translation tasks and has become the mainstream translation model. However, the calculation complexity of Transformer model is high, especially when dealing with long sequences, the inference efficiency becomes a key issue. In order to improve the inference efficiency of Transformer model, a method based on low-rank decomposition is adopted, and the calculation amount is significantly reduced by optimizing the key modules of the model. Firstly, this paper decomposes the self-attention mechanism in the Transformer model into low-rank, and decomposes the high-dimensional matrix into the product of several low-dimensional matrices by using matrix decomposition technology, thus reducing the computational complexity. Secondly, in the process of model training, pruning technology is combined to remove redundant parameters, so as to further reduce the calculation amount and model scale. On the premise of ensuring the translation quality, this paper carries out experimental verification on large-scale data sets, and the results show that this method can greatly improve the inference speed of the model while maintaining the translation performance. For short sentences with less than 100 words, the inference time of the original model is from 50 milliseconds to 140 milliseconds, and the inference time of the low-rank decomposition model is shortened to 40-100 milliseconds, which is a significant improvement for the real-time translation system. The experimental results show that the low-rank decomposition method not only significantly improves the inference efficiency of the model, but also effectively reduces the consumption of computing resources, which provides a more economical and efficient choice for practical application. © 2024 IEEE.

关键词： Computer aided language translation

来源：评论

学校读者我要写书评

暂无评论

Fusion strategy-driven dung beetle optimization algorithm

Fusion strategy-driven dung beetle optimization algorithm

引用

2024 International Conference on Algorithms, High Performance Computing, and Artificial Intelligence, AHPCAI 2024

作者： Huang, Xueyu Yi, Jinwei School of Software Engineering Jiangxi University of Science and Technology Jiangxi Nanchang330000 China

ISBN: (纸本)9781510685840

Considering the challenges posed by the dung beetle optimization algorithm, which often gravitates towards local optimality and disproportionate global exploration and local development potentials, a fusion strategy driven dung beetle optimization algorithm (FSDDBO) is proposed. Firstly, the piecewise linear chaotic map (PWLCM) is used to initialize the population, so that the dung beetle population can better traverse the entire solution space;secondly, an improved spiral search strategy is added in the dung beetle breeding stage to accelerate the convergence speed and improve individual diversity;finally, a competitive reverse learning strategy is added in the dung beetle stealing stage to increase the randomness of the dung beetle and update the existing worldwide most unfavorable solution. In 11 benchmark function tests and comparisons with other algorithms, the method proposed in this research improves the algorithm's capacity to leap beyond its local optimum, ensuring strong robustness and precision in optimization, achieving good results. © 2024 SPIE.

关键词： Piecewise linear techniques

来源：评论

学校读者我要写书评

暂无评论

Generative AI — The End of Systematic Reviews in PhD Projects?

引用

ACM Inroads 2024年第2期15卷 48-50页

作者： Zamani, Sanaz Sinha, Roopak Department of Computer Science and Software Engineering Auckland University of Technology Auckland New Zealand School of Information Technology Deakin University Melbourne Australia

In this opinion piece, we question the efficacy of students conducting systematic reviews (SRs) at the very start of their PhDs, especially now that we are riding, or drowning in, the Generative AI wave. How would the ubiquitous availability of Generative AI tools shape PhD research and supervision, and how can we capitalize on such disruptions? We report our reflections about early-stage PhD research from 10+ PhD supervisions completed by Sinha, and a recently started PhD project with Zamani as the student. We conducted a systematic literature review following traditional methods for the first four months and then revisited the task using ChatGPT and other related tools. We opine that while SRs still have value for new students, Generative AI must be incorporated, with care, to significantly enhance student learning, reinvest time savings into other, often-ignored skills, and improve the quality (breadth and depth) of SRs. © 2024 Association for Computing Machinery. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Ontology Evaluation Tools: Current and Future Research

引用

Recent Advances in Computer Science and Communications 2023年第6期16卷 2-15页

作者： Debnath, Narayan C. Patel, Archana Department of Software Engineering School of Computing and Information Technology Eastern International University Thu Dau Mot Viet Nam

In recent years, the increasing interest in ontologies resulted in the developing and publishing of many ontologies in the same or different domains. When users try to reuse the existing ontologies in their applications, they may encounter problems determining the most appropriate ontology based on their needs. Ontology evaluation is a process that helps users to examine the quality of the ontology based on different attributes. Many accessible and usable tools for ontology evaluation have been studied in the literature. However, finding an efficient ontology evaluation tool, following ontology specifications and their requirements (advantages/disadvantages), is still missing, limiting the researchers from determining possible future research. This paper aims to help new researchers and practitioners identify appropriate ontology evaluation tools based on their requirements and provide guidelines for future research directions on the same topic. This paper provides a detailed description of the different types of ontologies and classifies the available ontology evaluation tools into two categories, namely domain dependant ontology evaluation tools and domain-independent ontology evaluation tools. © 2023 Bentham Science Publishers.

关键词： Ontology

来源：评论

学校读者我要写书评

暂无评论

Research on Anomaly Detection in Time-Series Streaming Data 3

Research on Anomaly Detection in Time-Series Streaming Data

引用

3rd International Conference on Computer Science, Electronic Information engineering and Intelligent Control technology, CEI 2023

作者： Gong, Luyang School of Software Engineering Beijing University of Technology Beijing China

ISBN: (纸本)9798350372052

The detection of anomalies in streaming data is crucial in enterprise operations, employing statistical and machine learning methods to identify irregularities. This enhances service stability and reduces operational complexity. However, existing monitoring platforms face challenges such as complex rule configurations and a lack of unified standards. To address these issues, this paper proposes a container performance anomaly detection module that combines statistical and machine learning approaches. This module, through steps including streaming data collection, data preprocessing, anomaly detection models, and monitoring alerts, achieves rapid detection and accurate identification of container performance anomalies. In comparison to traditional monitoring platforms, this module not only saves manpower but also demonstrates higher accuracy. © 2023 IEEE.

关键词： Time series

来源：评论

学校读者我要写书评

暂无评论

Based on the Conformer End-to-End Chinese Speech Recognition Method 4

Based on the Conformer End-to-End Chinese Speech Recognition...

引用

4th International Conference on Electronic Information engineering and Computer Science, EIECS 2024

作者： Feng, Haiwen Shi, Yunxiao School of Software Engineering Shenyang University of Technology Shenyang Key Laboratory of Intelligent Manufacturing of High-end Industrial Equipment Liaoning Shenyang China School of Software Engineering Shenyang University of Technology ShenganpLiaoning China

ISBN: (纸本)9798331531409

Aiming at the problem that the acoustic input network of Conformer encoder is insufficient for Fbank speech feature extraction, an end-to-end speech recognition modeling method of DM-Conformer is proposed. Firstly, the Dilated convolution is utilized in the downsampling module to increase the receptive field, which captures more speech feature information without adding extra model parameters, while ensuring that the height and width of the original input feature map remain unchanged. Then, the Mish activation function is introduced to alleviate the problem of slow convergence of the ReLU activation function due to its negative value. Finally, experiments are carried out on the public dataset AISHELL-1, and the speech recognition model with the improved downsampling module proposed in this paper reduces the word error rate by 5.6% compared with the Conformer baseline model on the test set, and the model finally achieves a word error rate of 6.15%. © 2024 IEEE.

关键词： Speech enhancement

来源：评论

学校读者我要写书评

暂无评论

Masked hybrid attention with Laplacian query fusion and tripartite sequence matching for medical image segmentation

引用

Neural Computing and Applications 2025年第8期37卷 5891-5911页

作者： Ekong, Favour Yu, Yongbin Patamia, Rutherford Agbeshi Sarpong, Kwabena Ukwuoma, Chiagoziem C. Wang, Xiangxiang Ukot, Akpanika Robert Cai, Jingye School of Information and Software Engineering University of Electronic Science and Technology of China Sichuan Chengdu611731 China College of Nuclear Technology and Automation Engineering Chengdu University of Technology Sichuan Chengdu610059 China Sichuan Engineering Technology Research Center for Industrial Internet Intelligent Monitoring and Application Chengdu University of Technology Sichuan Chengdu610059 China

Medical image segmentation is pivotal in computer-aided diagnosis systems, demanding high precision and contextual understanding. Vision Transformer-based approaches have gained much attention recently due to their excellent performance and ability to capture long-range dependencies in medical images. However, research shows they suffer from inadequate multi-scale feature integration, poor object localization, and inconsistent mask predictions, leading to sub-optimal segmentation performance. This paper addresses these challenges by redefining semantic medical image segmentation through learnable object queries within an enhanced transformer framework with a masked hybrid attention querying mechanism, optimizing multi-scale feature fusion, object localization, and instance-specific segmentation. First, this study presents a novel transformer-based masked hybrid attention mechanism using Laplacian query fusion on learnable query features and incorporating a novel tripartite sequence matching technique as part of the enhanced decoder block to improve the consistency of mask predictions and optimize decoder queries. The designed hybrid multi-head self- and cross-attention mechanisms aim to selectively integrate multi-scale features, ensuring optimal feature combinations for precise segmentation. Secondly, multiple class tokens are incorporated to improve object localization and capture class-specific characteristics within the transformer framework, leveraging the transformer decoders’ ability to learn distinct instance representations. Experimental results and extensive ablation studies demonstrate the effectiveness of the proposed approach on three publicly available datasets, obtaining better segmentation results compared to various state-of-the-art approaches using various evaluation metrics. Specifically, the proposed model achieves a Dice Score of 95.25%, 92.75%, and 85.25% on LUNA, ISIC, and DRIVE datasets, respectively. © The Author(s), under exclusive licence

关键词： Laplace transforms

来源：评论

学校读者我要写书评

暂无评论

Exploration and Practice of Deep Learning Classroom Teaching of 'Computer Network' Course Based on the Integration of Specialization and Innovation 5

Exploration and Practice of Deep Learning Classroom Teaching...

引用

5th Asia Conference on Information engineering, ACIE 2025

作者： Huang, Min Peng, ShaoWu Yuan, Hua School of Software Engineering South China University of Technology Guangzhou China School of Computer Science South China University of Technology Guangzhou China

ISBN: (纸本)9798331528409

In the era of rapid development of artificial intelligence technologies, traditional teaching models are unable to meet the employment needs of enterprises, and talent cultivation in universities faces more challenges, so teachers should consider how to better cultivate students' innovative abilities in classroom teaching and guiding them in practical applications. The paper focuses on teaching reform and practice of 'Computer Networks' course by deep learning classroom, which mainly includes three aspects: 1) Integrating ideological and political education and innovative education into the process of professional education of in the curriculum, by which to enhance students' professional innovation awareness;2) Inviting experts from enterprises to participate in classroom teaching and practical guidance to help students learn about the ability requirements and related positions in the IT enterprises for talents in the era of AI;3) Designing practice projects and guiding students to implement them by 'design thinking' methods, and also adopting an evaluation method that can reflect the innovative ability of students. Works of the paper can better reflect the support and deepening the teaching philosophy of studentcentered, and provide an attempt to improve teaching effectiveness and talent cultivation quality. © 2025 IEEE.

关键词： Cultivation

来源：评论

学校读者我要写书评

暂无评论

SparseCoder: Identifier-Aware Sparse Transformer for File-Level Code Summarization 31

SparseCoder: Identifier-Aware Sparse Transformer for File-Le...

引用

31st IEEE International Conference on software Analysis, Evolution and Reengineering, SANER 2024

作者： Wang, Yanlin Huang, Yanxian Guo, Daya Zhang, Hongyu Zheng, Zibin School of Software Engineering Sun Yat-sen University China School of Computer Science and Technology Sun Yat-sen University China School of Big Data and Software Engineering Chongqing University China

ISBN: (纸本)9798350330663

Code summarization aims to generate natural language descriptions of source code, facilitating programmers to understand and maintain it rapidly. While previous code summarization efforts have predominantly focused on method-level, this paper studies file-level code summarization, which can assist programmers in understanding and maintaining large source code projects. Unlike method-level code summarization, file-level code summarization typically involves long source code within a single file, which makes it challenging for Transformer-based models to understand the code semantics for the maximum input length of these models is difficult to set to a large number that can handle long code input well, due to the quadratic scaling of computational complexity with the input sequence length. To address this challenge, we propose SparseCoder, an identifier-aware sparse transformer for effectively handling long code sequences. Specifically, the SparseCoder employs a sliding window mechanism for self-attention to model short-term dependencies and leverages the structure message of code to capture long-term dependencies among source code identifiers by introducing two types of sparse attention patterns named global and identifier attention. To evaluate the performance of SparseCoder, we construct a new dataset FILE-CS for file-level code summarization in Python. Experimental results show that our SparseCoder model achieves state-of-the-art performance compared with other pre-trained models, including full self-attention and sparse models. Additionally, our model has low memory overhead and achieves comparable performance with models using full self-attention mechanism. Furthermore, we verify the generality of SparseCoder on other code understanding tasks, i.e., code clone detection and code search, and results show that our model outperforms baseline models in both tasks, demonstrating that our model can generate better code representations for various downstream tasks. Our

关键词： Semantics

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：