检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Xia, Xiaobo Lu, Pengqian Gong, Chen Han, Bo Yu, Jun Yu, Jun Liu, Tongliang The Sydney AI Center School of Computer Science Faculty of Engineering The University of Sydney DarlingtonNSW2008 Australia The Australian AI Institute Faculty of Engineering and IT The University of Technology Sydney BroadwayNSW2007 Australia The PCA Lab The Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing210094 China The Department of Computing Hong Kong Polytechnic University Hong Kong The Department of Computer Science Hong Kong Baptist University Hong Kong The School of Computer Science and Technology Hangzhou Dianzi University Hangzhou310018 China The Department of Automation University of Science and Technology of China Hefei230026 China

The sample selection approach is very popular in learning with noisy labels. As deep networks "learn pattern first", prior methods built on sample selection share a similar training procedure: the small-loss examples can be regarded as clean examples and used for helping generalization, while the large-loss examples are treated as mislabeled ones and excluded from network parameter updates. However, such a procedure is arguably debatable from two folds: (a) it does not consider the bad influence of noisy labels in selected small-loss examples;(b) it does not make good use of the discarded large-loss examples, which may be clean or have meaningful information for generalization. In this paper, we propose regularly truncated M-estimators (RTME) to address the above two issues simultaneously. Specifically, RTME can alternately switch modes between truncated M-estimators and original M-estimators. The former can adaptively select small-losses examples without knowing the noise rate and reduce the side-effects of noisy labels in them. The latter makes the possibly clean examples but with large losses involved to help generalization. Theoretically, we demonstrate that our strategies are label-noise-tolerant. Empirically, comprehensive experimental results show that our method can outperform multiple baselines and is robust to broad noise types and levels. The implementation is available at https://***/xiaoboxia/RTM_LNL. Copyright © 2023, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Hypervelocity impact induced shock acoustic emission waves for quantitative damage evaluation using in situ miniaturized piezoelectric sensor network

引用

Chinese Journal of Aeronautics 2019年第5期32卷 1059-1070页

作者： Menglong LIU Qiang WANG Qingming ZHANG Renrong LONG Fangsen CUI Zhongqing SU Institute of High Performance Computing A*STAR (Agency for Science Technology and Research) School of Automation Nanjing University of Posts and Telecommunications State Key Laboratory of Explosion Science and Technology Beijing Institute of Technology Institute of High Performance Computing (A*STAR Agency for Science Technology and Research) Department of Mechanical Engineering The Hong Kong Polytechnic University

Manmade debris and natural meteoroids, travelling in the Low Earth Orbit at a speed of several kilometers per second, pose a severe safety concern to the spacecraft in service through the HyperVelocity Impact(HVI). To address this issue, an investigation of shock Acoustic Emission(AE) waves induced by HVI to a downscaled two-layer Whipple shielding structure is performed,to realize a quantitative damage evaluation. Firstly a hybrid numerical model integrating smoothparticle hydrodynamics and finite element is built to obtain the wave response. The projectiles, with various impact velocities and directions, are modelled to impact the shielding structure with different thicknesses. Then experimental validation is carried out with built-in miniaturized piezoelectric sensors to in situ sense the HVI-induced AE waves. A quantitative agreement is obtained between numerical and experimental results, demonstrating the correctness of the hybrid model and facilitating the explanation of obtained AE signals in experiment. Based on the understanding of HVI-induced wave components, assessment of the damage severity, i.e., whether the outer shielding layer is perforated or not, is performed using the energy ratio between the regions of ‘‘high frequency" and ‘‘low frequency" in the acquired AE signals. Lastly, the direct-arrival fundamentalsymmetric wave mode is isolated from each sensing signal to be input into an enhanced delay-andsum algorithm, which visualizes HVI spots accurately and instantaneously with different sensor network configuration. All these works demonstrate the potential of quantitative, in situ, and real time HVI monitoring using miniaturized piezoelectric sensor network.

关键词： Acoustic Emission(AE) Damage detection Hypervelocity impact Piezoelectric sensor Whipple shielding assembly

来源：评论

学校读者我要写书评

暂无评论

Nonequilibrium dynamics of the localization-delocalization transition in the non-Hermitian Aubry-André model

引用

Physical Review B 2022年第1期106卷 014204-014204页

作者： Liang-Jun Zhai Guang-Yao Huang Shuai Yin Department of Physics Nanjing University Nanjing 210093 China The School of Mathematics and Physics Jiangsu University of Technology Changzhou 213001 China Institute for Quantum Information & State Key Laboratory of High Performance Computing College of Computer Science and Technology National University of Defense Technology Changsha 410073 China School of Physics Sun Yat-Sen University Guangzhou 510275 China

In this paper, we investigate the driven dynamics of the localization transition in the non-Hermitian Aubry-André model with the periodic boundary condition. Depending on the strength of the quasiperiodic potential λ, this model undergoes a localization-delocalization phase transition. We find that the localization length ξ satisfies ξ∼ɛ−ν with ɛ being the distance from the critical point and ν=1 being a universal critical exponent independent of the non-Hermitian parameter. In addition, from the finite-size scaling of the energy gap between the ground state and the first excited state, we determine the dynamic exponent z as z=2. The critical exponent of the inverse participation ratio for the nth eigenstate is also determined as s=0.1197. By changing ɛ linearly to cross the critical point, we find that the driven dynamics can be described by the Kibble-Zurek scaling (KZS). Moreover, we show that the KZS with the same set of the exponents can be generalized to the localization phase transitions in the excited states.

关键词： Anderson localization Localization Scaling methods

来源：评论

学校读者我要写书评

暂无评论

F2UE-NET: An Uncertainty-Driven Dynamic Expert Network for Long-Tail Retinal Disease Classification

F2UE-NET: An Uncertainty-Driven Dynamic Expert Network for L...

引用

IEEE International Symposium on Biomedical Imaging

作者： Junyong Zhao Tian Lin Meng Wang Liang Sun Haoyu Chen Daoqiang Zhang Huazhu Fu MOE Key Laboratory of Brain Computer Intelligence Technology College of Artificial Intelligence Nanjing University of Aeronautics and Astronautics Nanjing China Joint Shantou International Eye Center Shantou University and the Chinese University of Hong Kong Shantou Guangdong China Department of Ophthalmology Yong Loo Lin School of Medicine National University of Singapore Singapore Singapore Agency for Science Technology and Research (A*STAR) Institute of High Performance Computing (IHPC) Singapore Singapore

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

The automatic diagnosis of retinal diseases presents a significant challenge in medical image analysis, primarily due to the long-tail distribution in existing retinal datasets. This imbalance creates substantial difficulties in model training and performance. To address this issue, we propose an uncertainty-driven dynamic expert network for long-tail retinal disease classification, called $\mathrm{F}^{2}\text{UE}$ -Net. Specifically, we apply Low-Rank Adaptation (LoRA) technology to fine-tune the foundation model, enhancing its feature extraction capabilities and overall performance. Additionally, we introduce an uncertainty-driven dynamic expert participation strategy, which assigns appropriate weights to different categories during training, ensuring high accuracy even for rare, tail-end diseases. Experimental results on a long-tail fundus disease dataset demonstrate that $\mathrm{F}^{2}$ UE outperforms state-of-the-art methods in long-tail retinal disease classification.

关键词： Training Heavily-tailed distribution Image analysis Foundation models Biological system modeling Tail Retina Feature extraction Medical diagnostic imaging Diseases

来源：评论

学校读者我要写书评

暂无评论

IRLS: An Improved Reinforcement Learning Scheduler for high performance computing Systems

IRLS: An Improved Reinforcement Learning Scheduler for High ...

引用

International Conference on System science and engineering (ICSSE)

作者： Thanh Hoang Le Hai Luan Le Dinh Dat Ngo Tien Dat Bui Huu Tien Nam Thoai High Performance Computing Laboratory Advanced Institute of Interdisciplinary Science and Technology Faculty of Computer Science and Engineering Ho Chi Minh City University of Technology (HCMUT) 268 Ly Thuong Kiet Street District 10 Ho Chi Minh City Vietnam Vietnam National University Ho Chi Minh City Linh Trung Ward Thu Duc City Ho Chi Minh City Vietnam

Exploiting current high performance computing (HPC) systems is a critical task for resolving urgent worldwide problems. However, existing scheduling heuristics such as First Come First Served (FCFS) have limitations in dealing with the increasing complexity of computing systems and the dynamic nature of application workloads. Reinforcement learning (RL) has emerged as a promising approach to designing HPC schedulers that can learn to adapt to dynamic system configurations and workload conditions. However, existing RL-based schedulers often lack the ability to incorporate important identity features of jobs and do not consider user *** address these limitations, we propose an improvement to the latest Deep Reinforcement Learning Agent for Scheduling (DRAS) model, called Improved Reinforcement Learning Scheduler (IRLS). The IRLS model incorporates additional identity features in the state definition to recognize similarities between tasks from the same source and utilizes an empirical approach to perform job runtime prediction. Our experiments demonstrate that by using the IRLS model, we can significantly improve the performance of real-life HPC workloads, with improvements of up to 15.4% compared to the original DRAS model and 35.7% compared to FCFS.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Jacobi Collocation Method for the Fractional Ginzburg-Landau Differential Equation

引用

Advances in Applied Mathematics and Mechanics 2020年第1期12卷 57-86页

作者： Yin Yang Jianyong Tao Shangyou Zhang Petr V.Sivtsev Hunan Key Laboratory for Computation and Simulation in Science and Engineering Key Laboratory of Intelligent Computing&Information Processing of Ministry of EducationSchool of Mathematics and Computational ScienceXiangtan UniversityXiangtanHunan 411105China Department of Mathematical Sciences University of DelawareNewarkDE 19716USA International Scientific and Research Laboratory of Multiscale Model Reduction and High Performance Computing Ammosov North Eastern UniversityKulakovskogo677013YakutskRussia

In this paper,we design a collocation method to solve the fractional Ginzburg-Landau equation.A Jacobi collocation method is developed and implemented in two ***,we space-discretize the equation by the Jacobi-Gauss-Lobatto collocation(JGLC)method in one-and two-dimensional *** equation is then converted to a system of ordinary differential equations(ODEs)with the time variable based on *** second step applies the Jacobi-Gauss-Radau collocation(JGRC)method for the time ***,we give a theoretical proof of convergence of this Jacobi collocation method and some numerical results showing the proposed scheme is an effective and high-precision algorithm.

关键词： The fractional Ginzburg-Landau equation Jacobi collocation method convergence

来源：评论

学校读者我要写书评

暂无评论

Meta-Learning Based Infrared Ship Object Detection Model for Generalization to Unknown Domains

SSRN

引用

SSRN 2023年

作者： Feng, Hui Tang, Wei Xu, Haixiang Jiang, Chengxin Ge, Shuzhi Sam He, Jianhua Key Laboratory of High-Performance Ship Technology of Ministry of Education Wuhan University of Technology Wuhan430063 China School of Naval Architecture Ocean and Energy Power Engineering Wuhan University of Technology Wuhan430063 China Department of Electrical and Computer Engineering National University of Singapore Singapore117576 Singapore School of Computer Science and Electronic Engineering University of Essex CO4 3SQ United Kingdom

Infrared images exhibit considerable variations in probability distributions, stemming from the utilization of distinct infrared sensors and the influence of diverse environmental conditions. The variations pose great challenges for deep learning models to detect ship objects and adapt to unseen maritime environments. To address the domain shift problem, we propose an end-to-end infrared ship object detection model based on meta-learning neural network to improve domain adaptation for target domain where data is not available at training phase. Different from existing domain generalization methods, the novelty of our model lies in the effective exploition of meta-learning and domain adaptation, ensuring that the extracted domain-independent features are meaningful and domain-invariant at the semantic level. Firstly, a double gradient-based meta-learning algorithm is designed to solve the common optimal descent direction between different domains through two gradient updates in the inner and outer loops. The algorithm enables extraction of domain-invariant features from the pseudo-source and pseudo-target domain data. Secondly, a domain discriminator with dynamic-weighted gradient reversal layer (DWGRL) is designed to accurately classify domain-invariant features and provide additional global supervision information. Finally, a multi-scale feature aggregation method is proposed to improve the extraction of multi-scale domain-invariant features. It can effectively fuse local features at different scales and global features of targets. Extensive experimental results conducted in real nighttime water surface scenes demonstrate that the proposed model achieves very high detection accuracy on target domain data, even no target domain data was used during the training phase. Compared to the existing methods, our method not only improves the detection accuracy of infrared ships by 18%, but also exhibits the smallest standard deviation with a value of 0.93, indicating its su

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Towards Reliable Medical Image Segmentation by utilizing Evidential Calibrated Uncertainty

arXiv

引用

arXiv 2023年

作者： Zou, Ke Chen, Yidi Huang, Ling Yuan, Xuedong Shen, Xiaojing Wang, Meng Goh, Rick Siow Mong Liu, Yong Fu, Huazhu National Key Laboratory of Fundamental Science on Synthetic Vision Sichuan University Chengdu China College of Computer Science Sichuan University Chengdu China College of Mathematics Sichuan University Chengdu China Department of Radiology West China Hospital Sichuan University Chengdu China Saw Swee Hock School of Public Health National University of Singapore Singapore Institute of High Performance Computing Agency for Science Technology and Research Singapore

Medical image segmentation is critical for disease diagnosis and treatment assessment. However, concerns regarding the reliability of segmentation regions persist among clinicians, mainly attributed to the absence of confidence assessment, robustness, and calibration to accuracy. To address this, we introduce DEviS, an easily implementable foundational model that seamlessly integrates into various medical image segmentation networks. DEviS not only enhances the calibration and robustness of baseline segmentation accuracy but also provides high-efficiency uncertainty estimation for reliable predictions. By leveraging subjective logic theory, we explicitly model probability and uncertainty for the problem of medical image segmentation. Here, the Dirichlet distribution parameterizes the distribution of probabilities for different classes of the segmentation results. To generate calibrated predictions and uncertainty, we develop a trainable calibrated uncertainty penalty. Furthermore, DEviS incorporates an uncertainty-aware filtering module, which utilizes the metric of uncertainty-calibrated error to filter reliable data within the dataset. We conducted validation studies to assess both the accuracy and robustness of DEviS segmentation, along with evaluating the efficiency and reliability of uncertainty estimation. These evaluations were performed using publicly available datasets including ISIC2018, LiTS2017, and BraTS2019. Additionally, two potential clinical trials are being conducted at Johns Hopkins OCT, Duke-OCT-DME, and FIVES datasets to demonstrate their efficacy in filtering high-quality or out-of-distribution data. Our code has been released in https://***/Cocofeat/DEviS. © 2023, CC BY.

关键词： Calibration

来源：评论

学校读者我要写书评

暂无评论

Federated Learning for Recognizing Private Handwritten Numbers

Federated Learning for Recognizing Private Handwritten Numbe...

引用

2021 IEEE International Conference on Data science and computer Application, ICDSCA 2021

作者： Wang, Di Jiang, Yuchen Zhao, Yibo Chen, Panyang Li, Ang Pan, Kun Institution of Information Engineering Chinese Academy of Sciences State Key Laboratory of Information Security Beijing China School of Cyber Security University of Chinese Academy of Sciences Beijing China Computer Science and Technology College of Computer Science Chongqing University Chongqing China Zibo Shiyan High School International Division Zibo Shandong China Computer Science Department of Computer Science and Software Engineering Auburn University Auburn United States Shanghai Shangde Experimental School Shanghai China School of Computing University of Leeds Leeds United Kingdom

ISBN: (纸本)9781665440530

Federated learning (FL) is a new technology in the field of machine learning, which trains an algorithm across multiple decentralized edge devices or servers holding data samples, without exchanging their data samples. Horizontal Federated Learning (HFL) is a FL method focusing on using the same feature space across all users, which is suitable for situations where users' data feature dimensions overlap more. Because FL has the advantage of privacy protection in the field of manufacturing, we propose an HFL-based deep learning method named FedCNN, which realizes the digits handwritten recognition without violating the privacy or data sharing. Three different methods (local model, centralized model, and HFL-based FedCNN) are used on a public data set (MINIST) to compare the pros and cons of the solutions. The experimental results demonstrate that FedCNN scheme can simultaneously achieve practical privacy preservation and good model generalization performance with Homomorphic Encryption. © 2021 IEEE.

关键词： Data privacy

来源：评论

学校读者我要写书评

暂无评论

Robust Model Aggregation for Heterogeneous Federated Learning: Analysis and Optimizations

arXiv

引用

arXiv 2024年

作者： Shao, Yumeng Li, Jun Shi, Long Wei, Kang Ding, Ming Li, Qianmu Li, Zengxiang Chen, Wen Jin, Shi The School of Electronic and Optical Engineering Nanjing University of Science and Technology Nanjing210094 China Data61 CSIRO Sydney Australia The Digital Economy Research Institute Nanjing University of Science and Technology Jiangsu Nanjing210094 China The Institute of High Performance Computing A*STAR Singapore The Digital Research Institute ENN Group China The Department of Electronic Engineering Shanghai Jiao Tong University Shanghai200240 China The National Mobile Communications Research Laboratory Southeast University Nanjing210096 China

Conventional synchronous federated learning (SFL) frameworks suffer from performance degradation in heterogeneous systems due to imbalanced local data size and diverse computing power on the client side. To address this problem, asynchronous FL (AFL) and semi-asynchronous FL have been proposed to recover the performance loss by allowing asynchronous aggregation. However, asynchronous aggregation incurs a new problem of inconsistency between local updates and global updates. Motivated by the issues of conventional SFL and AFL, we first propose a time-driven SFL (T-SFL) framework for heterogeneous systems. The core idea of T-SFL is that the server aggregates the models from different clients, each with varying numbers of iterations, at regular time intervals. To evaluate the learning performance of T-SFL, we provide an upper bound on the global loss function. Further, we optimize the aggregation weights to minimize the developed upper bound. Then, we develop a discriminative model selection (DMS) algorithm that removes local models from clients whose number of iterations falls below a predetermined threshold. In particular, this algorithm ensures that each client’s aggregation weight accurately reflects its true contribution to the global model update, thereby improving the efficiency and robustness of the system. To validate the effectiveness of T-SFL with the DMS algorithm, we conduct extensive experiments using several popular datasets including MNIST, Cifar-10, Fashion-MNIST, and SVHN. The experimental results demonstrate that T-SFL with the DMS algorithm can reduce the latency of conventional SFL by 50%, while achieving an average 3% improvement in learning accuracy over state-of-the-art AFL algorithms. Copyright © 2024, The Authors. All rights reserved.

关键词： Learning algorithms

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：