检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Wei, Qi Feng, Lei Wang, Haobo An, Bo School of Computer Science and Engineering Nanyang Technological University Singapore School of Software Zhejiang University China

Learning with noisy labels aims to ensure model generalization given a label-corrupted training set. The sample selection strategy achieves promising performance by selecting a label-reliable subset for model training. In this paper, we empirically reveal that existing sample selection methods suffer from both data and training bias that are represented as imbalanced selected sets and accumulation errors in practice, respectively. However, only the training bias was handled in previous studies. To address this limitation, we propose a noIse-Tolerant Expert Model (ITEM) for debiased learning in sample selection. Specifically, to mitigate the training bias, we design a robust network architecture that integrates with multiple experts. Compared with the prevailing double-branch network, our network exhibits better performance of selection and prediction by ensembling these experts while training with fewer parameters. Meanwhile, to mitigate the data bias, we propose a mixed sampling strategy based on two weight-based data samplers. By training on the mixture of two class-discriminative mini-batches, the model mitigates the effect of the imbalanced training set while avoiding sparse representations that are easily caused by sampling strategies. Extensive experiments and analyses demonstrate the effectiveness of ITEM. Our code is available at this url ITEM. Copyright © 2024, The Authors. All rights reserved.

关键词： Network architecture

来源：评论

学校读者我要写书评

暂无评论

LLAMAFACTORY: Unified Efficient Fine-Tuning of 100+ Language Models

arXiv

引用

arXiv 2024年

作者： Zheng, Yaowei Zhang, Richong Zhang, Junhao Ye, Yanhan Luo, Zheyan Feng, Zhangchi Ma, Yongqiang School of Computer Science and Engineering Beihang University China School of Software and Microelectronics Peking University China

Efficient fine-tuning is vital for adapting large language models (LLMs) to downstream tasks. However, it requires non-trivial efforts to implement these methods on different models. We present LLAMAFACTORY, a unified framework that integrates a suite of cutting-edge efficient training methods. It provides a solution for flexibly customizing the fine-tuning of 100+ LLMs without the need for coding through the built-in web UI LLAMABOARD. We empirically validate the efficiency and effectiveness of our framework on language modeling and text generation tasks. It has been released at https://***/hiyouga/LLaMA-Factory and received over 25,000 stars and 3,000 forks. Copyright © 2024, The Authors. All rights reserved.

关键词： Unified Modeling Language

来源：评论

学校读者我要写书评

暂无评论

Reliable Routing and Scheduling in Time Sensitive Networks based on Reinforcement Learning

引用

IEEE Transactions on Network science and engineering 2025年

作者： Cheng, Hao Yang, Lei Zhang, Qingfeng Zhu, Weiping South China University of Technology School of Software Engineering Guangzhou510006 China Wuhan University School of Computer Science Wuhan430072 China

Time Sensitive Network (TSN) provides strict low latency and bounded jitter requirements for applications such as industrial systems, autonomous driving, etc. One of the important problems in TSN is to achieve high reliability and low latency by effectively routing and scheduling time-sensitive data flows. Existing work applies heuristic or integer programming to address flow routing and scheduling, yet often fail to achieve optimal solutions quickly. In this paper, we propose a new Reinforcement Learning (RL) based approach for routing and scheduling of redundant data flows, aiming to achieve load balancing on the network links as well as meeting the reliability and delay constraints. Our approach first leverages a simple heuristic algorithm to decide the redundant path candidate set, and then incorporates Proximal Policy Optimization (PPO) method to choose the most suitable multi-routing flows from the candidates, which can be aware of the network status dynamically to reduce the load on the bottleneck link of the network. On this basis, we further retrain the RL model by fine-tuning to adapt to the online environment. The simulation results show that our proposed solution outperforms the benchmark algorithms in terms of the degree of network balance by 38.7% in offline network environments and in terms of average delay by 14.0% in online network environments. © 2025 IEEE. All rights reserved.

关键词： Delay tolerant networks

来源：评论

学校读者我要写书评

暂无评论

Slicing Input Features to Accelerate Deep Learning: A Case Study with Graph Neural Networks

arXiv

引用

arXiv 2024年

作者： Xu, Zhengjia Lyu, Dingyang Zhang, Jinghui College of Software Engineering Southeast University China School of Computer Science and Engineering Southeast University China

As graphs grow larger, full-batch GNN training becomes hard for single GPU memory. Therefore, to enhance the scalability of GNN training, some studies have proposed sampling-based mini-batch training and distributed graph learning. However, these methods still have drawbacks, such as performance degradation and heavy communication. This paper introduces SliceGCN, a feature-sliced distributed large-scale graph learning method. SliceGCN slices the node features, with each computing device, i.e., GPU, handling partial features. After each GPU processes its share, partial representations are obtained and concatenated to form complete representations, enabling a single GPU's memory to handle the entire graph structure. This aims to avoid the accuracy loss typically associated with mini-batch training (due to incomplete graph structures) and to reduce inter-GPU communication during message passing (the forward propagation process of GNNs). To study and mitigate potential accuracy reductions due to slicing features, this paper proposes feature fusion and slice encoding. Experiments were conducted on six node classification datasets, yielding some interesting analytical results. These results indicate that while SliceGCN does not enhance efficiency on smaller datasets, it does improve efficiency on larger datasets. Additionally, we found that SliceGCN and its variants have better convergence, feature fusion and slice encoding can make training more stable, reduce accuracy fluctuations, and this study also discovered that the design of SliceGCN has a potentially parameter-efficient nature. Copyright © 2024, The Authors. All rights reserved.

关键词： Encoding (symbols)

来源：评论

学校读者我要写书评

暂无评论

Guest Editorial: Special Issue on Human-Machine Fusion Decision-Making for Emergency Handling

引用

IEEE Transactions on Automation science and engineering 2025年 22卷 4427-4433页

作者： Wu, Edmond Q. Li, Jianqiang Chen, Guimin Yuce, Mehmet R. Islam, Shafiqul Ser, Javier Del Yu, Hui Liu, Peter Xiaoping Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai200240 China College of Computer and Software Engineering Shenzhen University Shenzhen518060 China State Key Laboratory for Manufacturing Xi'an Jiaotong University Xi'an710049 China Department of Electrical and Computer Systems Engineering Monash University MelbourneVIC3800 Australia Xavier University of Louisiana New OrleansLA70125 United States Tecnalia Institute Bizkaia48160 Spain School of Creative Technologies University of Portsmouth PortsmouthPO1 2UP United Kingdom Department of Systems and Computer Engineering Carleton University OttawaONKIS 5B6 Canada

来源：评论

学校读者我要写书评

暂无评论

Image Reconstruction Improvement of Variable Coded Aperture using Deep Learning Method for Gamma and Lensless Imaging Applications

Image Reconstruction Improvement of Variable Coded Aperture ...

引用

2023 Conference on Lasers and Electro-Optics Europe and European Quantum Electronics Conference, CLEO/Europe-EQEC 2023

作者： Schwarz, Ariel Shemer, Amir Danan, Eliezer Cohen, Noa E. Danan, Yossef Department of Electrical and Electronics Engineering Azrieli College of Engineering Jerusalem9103501 Israel Faculty of Engineering Bar-Ilan University Ramat-Gan5290002 Israel School of Software Engineering and Computer Science Azrieli College of Engineering Jerusalem9103501 Israel

ISBN: (纸本)9798350345995

In gamma ray imaging for nuclear medicine, coded aperture is used to improve sensitivity. one of the main reconstructing methods is inverse filtering (deconvolution), where the recorded image is cross-correlated with periodic inverse filter of coded array. The reconstruction is free of coding noise for arbitrary array. However, amplification of quantum noise affect the reconstructed image. Although it is improved by Wiener filtering, the major problem is small terms in the spectral distribution of coded masks. Statistically and experimentally pinhole arrays have at least one term which is zero, resulting unacceptably noisy reconstruction. In our previous research we presented new approach of variable coded aperture (VCA) design for far and near field imaging applications [1-4]. The imaging system is based on time multiplexing method using variable multi pinhole array. The unique variable design enables to overcome the spatial frequencies cutoff and small terms in the Fourier transform exists in static multi pinhole array. The overall pinholes positions are designed to avoid spatial frequencies loss. However, traces of duplications can still be detected in reconstruction using Wiener filtering. Furthermore, since coded aperture blocks most of the photons that enter the detector, the dynamic range of the image is limited, thus leading to low contrast image and inadequate colors gamut. © 2023 IEEE.

关键词： Quantum noise

来源：评论

学校读者我要写书评

暂无评论

DCQNet: Collaborative Camouflaged Object Detection Using Cross-Sample and Cross-Scale Network

DCQNet: Collaborative Camouflaged Object Detection Using Cro...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Panrui Tang Zuping Zhang Yubin Sheng Bo Huang Yao Xiao Lin Shen School of Computer Science and Engineering Central South University School of Software Xinjiang University Changsha China

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Camouflaged object detection (COD) aims to identify objects that blend into the surrounding backgrounds, which has been a hot topic in recent years, with many different optimization strategies being explored. Among these, collaborative detection, a recently proposed solution for COD, has shown outstanding performance. However, current collaborative detection methods adopt a "one-to-many" pattern, failing to fully leverage the advantages of collaborative detection. Moreover, existing approaches to disguised target detection overlook the importance of high-level features. To address these issues, we propose a novel dualcross query network (DCQNet). It effectively capitalizes on the commonalities among objects and the directive role of high-level features in enhancing low-level features. Specifically, we designed a cross-sample query module and a cross-scale query module to collaboratively locate the object and guide low-level features, respectively. Extensive experimental results demonstrate that DCQNet outperforms state-of-the-art (SOTA) methods on the CoCOD8K dataset.

关键词： Representation learning Accuracy Semantics Neural networks Collaboration Object detection Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Usability Evaluation of Kids' Learning Apps 2

Usability Evaluation of Kids' Learning Apps

引用

2nd International Conference on Business Analytics for Technology and Security, ICBATS 2023

作者： Ibrahim, Amer Al-Rajab, Murad Hamid, Khalid Aqeel, Muhamamd Muneer, Salman Parveen, Mehvish Saleem, Maryam American University in the Emirates College of Computer and Information Technology Dubai United Arab Emirates Abu Dhabi University Computer Scienc and IT College of Engineering United Arab Emirates Superior University Department of Computer Science Lahore Pakistan Superior University Department of Software Engineering Lahore Pakistan NCBA&E School of Computer Science Lahore Pakistan Applied Science Private University Applied Science Research Center Amman11937 Jordan NCBAE School of Computer Science Lahore Pakistan

ISBN: (纸本)9798350335644

The main objective of learning applications is to use modern electronic technologies to impart knowledge, communicate knowledge, and Direct learning models with their learning activities in a timely and effective manner. The preponderance of the apps were designed to up-skill the kids in the fundamentals of letters and numbers. Generally, they were grounded and practice skills in nature, based on a lack of abilities in critical thinking, favoring parrot-fashion skills, and contributing to a deep conceptual comprehension of more specific concepts are unable. By using appropriate evaluation methodologies, the purposed study observes essential characteristics and aspects that affect the usability of interfaces. An evaluation of the app's features will figure out to ensure user contentment. When a kid interacts with a learning interface, the ease of use provided by the interface is validated. Because a large number of the apps were designed to train kids' fundamentals of alphabetic and numeric. The focus of this research is to see whether self-proclaimed learning apps for kids are built according to developmentally appropriate standards to help kids in formal and informal learning settings develop socially, emotionally, and academically. © 2023 IEEE.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Multi Task-Based Facial Expression Synthesis with Supervision Learning and Feature Disentanglement of Image Style

Multi Task-Based Facial Expression Synthesis with Supervisio...

引用

IEEE International Conference on Image Processing

作者： Wenya Lu Zhibin Peng Cheng Luo Weicheng Xie Jiajun Wen Zhihui Lai Linlin Shen Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University

Image-to-Image synthesis paradigms have been widely used for facial expression synthesis. However, current generators are apt to either produce artifacts for largely posed and non-aligned faces or unduly change the identity information like AdaIN-based generator. In this work, we suggest to use image style feature to surrogate the expression cues in the generator, and propose a multi-task learning paradigm to explore this style information via the supervision learning and feature disentanglement. While the supervision learning can make the encoded style specifically represent the expression cues and enable the generator to produce correct expression, the feature disentanglement of content and style cues enables the generator to better preserve the identity information in expression synthesis. Experimental results show that the proposed algorithm can well reduce the artifacts for the synthesis of posed and non-aligned expressions, and achieves competitive performances in terms of FID, PNSR and classification accuracy, compared with four publicly available GANs. The code and pre-trained models are available at https://***/lumanxi236/MTSS.

关键词：

来源：评论

学校读者我要写书评

暂无评论

E2HQV: High-Quality Video Generation from Event Camera via Theory-Inspired Model-Aided Deep Learning

arXiv

引用

arXiv 2024年

作者： Qu, Qiang Shen, Yiran Chen, Xiaoming Chung, Yuk Ying Liu, Tongliang School of Computer Science The University of Sydney Australia School of Software Shandong University China School of Computer Science and Engineering Beijing Technology and Business University China

The bio-inspired event cameras or dynamic vision sensors are capable of asynchronously capturing per-pixel brightness changes (called event-streams) in high temporal resolution and high dynamic range. However, the non-structural spatial-temporal event-streams make it challenging for providing intuitive visualization with rich semantic information for human vision. It calls for events-to-video (E2V) solutions which take event-streams as input and generate high quality video frames for intuitive visualization. However, current solutions are predominantly data-driven without considering the prior knowledge of the underlying statistics relating event-streams and video frames. It highly relies on the non-linearity and generalization capability of the deep neural networks, thus, is struggling on reconstructing detailed textures when the scenes are complex. In this work, we propose E2HQV, a novel E2V paradigm designed to produce high-quality video frames from events. This approach leverages a model-aided deep learning framework, underpinned by a theory-inspired E2V model, which is meticulously derived from the fundamental imaging principles of event cameras. To deal with the issue of state-reset in the recurrent components of E2HQV, we also design a temporal shift embedding module to further improve the quality of the video frames. Comprehensive evaluations on the real world event camera datasets validate our approach, with E2HQV, notably outperforming state-of-the-art approaches, e.g., surpassing the second best by over 40% for some evaluation metrics. © 2024, CC BY.

关键词： Cameras

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：