检索结果-内蒙古大学图书馆

arXiv 2020年

作者： Zhu, Boqing Xu, Kele Kong, Qiuqiang Wang, Huaimin Peng, Yuxing Science and Technology on Parallel and Distributed Laboratory National University of Defense Technology Changsha410073 China Centre for Vision Speech and Signal Processing University of Surrey GuildfordGU2 7XH United Kingdom

—High quality labeled datasets have allowed deep learning to achieve impressive results on many sound analysis tasks. Yet, it is labor-intensive to accurately annotate large amount of audio data, and the dataset may contain noisy labels in the practical settings. Meanwhile, the deep neural networks are susceptive to those incorrect labeled data because of their outstanding memorization ability. In this paper, we present a novel framework, named CrossFilter, to combat the noisy labels problem for audio tagging. Multiple representations (such as, Logmel and MFCC) are used as the input of our framework for providing more complementary information of the audio. Then, though the cooperation and interaction of two neural networks, we divide the dataset into curated and noisy subsets by incrementally pick out the possibly correctly labeled data from the noisy data. Moreover, our approach leverages the multitask learning on curated and noisy subsets with different loss function to fully utilize the entire dataset. The noisy-robust loss function is employed to alleviate the adverse effects of incorrect labels. On both the audio tagging datasets FSDKaggle2018 and FSDKaggle2019, empirical results demonstrate the performance improvement compared with other competing approaches. On FSDKaggle2018 dataset, our method achieves state-of-the-art performance and even surpasses the ensemble models. Copyright © 2020, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

NUMA-aware FFT-based convolution on ARMv8 many-core CPUs

arXiv

引用

arXiv 2021年

作者： Huang, Xiandong Wang, Qinglin Lu, Shuyu Hao, Ruochen Mei, Songzhu Liu, Jie Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha410073 China School of Computer Science National University of Defense Technology Changsha410073 China Department of Biomedical Informatics University of Pittsburgh Pittsburgh United States

Convolutional Neural Networks (CNNs), one of the most representative algorithms of deep learning, are widely used in various artificial intelligence applications. Convolution operations often take most of the computational overhead of CNNs. The FFT-based algorithm can improve the efficiency of convolution by reducing its algorithm complexity, there are a lot of works about the high-performance implementation of FFT-based convolution on many-core CPUs. However, there is no optimization for the non-uniform memory access (NUMA) characteristics in manycore CPUs. In this paper, we present a NUMA-aware FFT-based convolution implementation on ARMv8 many-core CPUs with NUMA architectures. The implementation can reduce a number of remote memory access through the data reordering of FFT transformations and the three-level parallelization of the complex matrix multiplication. The experiment results on a ARMv8 many-core CPU with NUMA architectures demonstrate that our NUMA-aware implementation has much better performance than the state-of-the-art work in most cases. Copyright © 2021, The Authors. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Big data collaborative artificial intelligence and high-performance computing to drive physical design of fusion

Big data collaborative artificial intelligence and high-perf...

引用

2022 International Conference on Cloud Computing, Big Data Applications and Software Engineering, CBASE 2022

作者： Liu, Pan Liu, Gongjie Liu, Jie Liu, Chencheng Xiong, Wenbin Tang, Zhangcun Zhou, Yinzhao Wang, Yonggang Yuan, Zhouming Beijng Paratera Technology Corp. Ltd Beijing100000 China National University of Defense Technology Science and Technology on Parallel and Distributed Processing Laboratory Changsha410073 China National University of Defense Technology Laboratory of Software Engineering for Complex Systems Changsha410073 China Central South University of Forestry and Technology Changsha410000 China Nuclear and Radiation Safety Center Mee Beijing100000 China University of Electronic Science and Technology of China Chengdu610000 China Tianfu Innovation Energy Establishment Chengdu610000 China

ISBN: (纸本)9798350396393

Controlled thermonuclear fusion has always been a dream pursued by mankind. However, the physical processes of controlled thermonuclear fusion are complex, requiring numerical simulations with high performance computing, and the amount of data generated by the physical processes on spatial, temporal and temperature scales is too large to be captured, managed, processed and collated in a reasonable time frame by mainstream software tools to achieve more aggressive fusion physical design. The data are too large to be captured, managed, processed, and collated into more aggressive targets for fusion physical design in a reasonable time by mainstream software tools. At the same time, the failure of fusion ignition can be caused by the distortion of various key physical quantities, and only by decomposing the process step by step and clarifying the changes of key physical quantities in the fusion physics process, can an effective mechanism be formed to prevent the distortion of key physical quantities from causing ignition failure in experimental physics. Big data in collaboration with artificial intelligence and high performance computing to drive the physical design of fusion is a novel avenue. By data acquisition with and pre-processing, this allows the creation of small sample libraries and deep learning. With supervised learning function convergence, incorporating solid/fluid computational methods, further network layering and cell expansion, the parameters of the physical model conforming to Lawson's criterion will become experimental physical parameters, and we find the new approach in Data capacity, Model combination approach, Type of material, Calculation speed, Optimisation of design iteration times, etc. are superior to the traditional approach. © 2022 IEEE.

关键词： Big data

来源：评论

学校读者我要写书评

暂无评论

An Effective Design to Improve the Efficiency of DPUs on FPGA

An Effective Design to Improve the Efficiency of DPUs on FPG...

引用

International Conference on parallel and distributed Systems (ICPADS)

作者： Yutian Lei Qingyong Deng Saiqin Long Shaohui Liu Sangyoon Oh The Key Laboratory of Hunan Province for Internet of Things and Information Security The School of Computer Science Science and Technology on Parallel and Distributed Processing Laboratory Xiangtan University Xiangtan China The Artificial Intelligence Center PengCheng Laboratory shenzhen China Ajou University Suwon-si Republic of Korea

The following topics are dealt with: cloud computing; learning (artificial intelligence); resource allocation; security of data; mobile computing; neural nets; optimisation; parallel processing; scheduling; and storag... 详细信息

ISBN: (数字)9781728190747

ISBN: (纸本)9781728183824

关键词： Deep learning Schedules Multicore processing Logic gates Throughput Acceleration Field programmable gate arrays

来源：评论

学校读者我要写书评

暂无评论

Excitement Surfeited Turns to Errors: Deep Learning Testing Framework Based on Excitable Neurons

arXiv

引用

arXiv 2022年

作者： Jin, Haibo Chen, Ruoxi Zheng, Haibin Chen, Jinyin Cheng, Yao Yu, Yue Liu, Xianglong College of Information Engineering Zhejiang University of Technology Hangzhou China Institute of Cyberspace Security Zhejiang University of Technology Hangzhou China Huawei International Singapore National Laboratory for Parallel and Distributed Processing College of Computer National University of Defense Technology Changsha China State Key Laboratory of Software Development Environment Beihang University Beijing China

Despite impressive capabilities and outstanding performance, deep neural networks (DNNs) have captured increasing public concern about their security problems, due to their frequently occurred erroneous behaviors. Therefore, it is necessary to conduct a systematical testing for DNNs before they are deployed to real-world applications. Existing testing methods have provided fine-grained metrics based on neuron coverage and proposed various approaches to improve such metrics. However, it has been gradually realized that a higher neuron coverage does not necessarily represent better capabilities in identifying defects that lead to errors. Besides, coverage-guided methods cannot hunt errors due to faulty training procedure. So the robustness improvement of DNNs via retraining by these testing examples are unsatisfactory. To address this challenge, we introduce the concept of excitable neurons based on Shapley value and design a novel white-box testing framework for DNNs, namely DeepSensor. It is motivated by our observation that neurons with larger responsibility towards model loss changes due to small perturbations are more likely related to incorrect corner cases due to potential defects. By maximizing the number of excitable neurons concerning various wrong behaviors of models, DeepSensor can generate testing examples that effectively trigger more errors due to adversarial inputs, polluted data and incomplete training. Extensive experiments implemented on both image classification models and speaker recognition models have demonstrated the superiority of DeepSensor. Compared with the state-of-the-art testing approaches, DeepSensor can find more test errors due to adversarial inputs (∼ ×1.2), polluted data (∼ ×5) and incompletely-trained DNNs (∼ ×1.3). Additionally, it can help DNNs build larger l2-norm robustness bound (∼ ×3) via retraining according to CLEVER's certification. We further provide interpretable proofs for effectiveness of DeepSensor via excitable neuro

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Big Data-aided Study of the Physical Process of Volume Ignition 6

Big Data-aided Study of the Physical Process of Volume Ignit...

引用

2022 6th International Conference on Mechanics, Mathematics and Applied Physics, ICMMAP 2022

作者： Liu, Pan Zhou, Yinzhao Xiong, Wenbin Liu, Chencheng Liu, Jie Liu, Gongjie Tang, Zhangchun Gao, Qiang Tianfu Innovation Energy Establishment Chengdu610000 China Nuclear and Radiation Safety Center MEE Beijing100000 China Central South University of Forestry and Technology Changsha410000 China Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha410073 China Beijng PARATERA Technology Corp. Ltd Beijing100000 China University of Electronic Science and Technology of China Chengdu610000 China

Volume ignition is a method of igniting a fuel as a whole by simultaneously achieving ignition conditions throughout the fuel zone. The basic criterion for ignition is that the thermonuclear energy is greater than the energy leakage at the fuel boundary, resulting in self-sustaining heating and deep combustion. Deuterium-tritium fuels are wrapped in medium to high Z media to reduce radiative leakage and achieve lower-temperature holistic ignition and non-equilibrium combustion, ultimately allowing the fuel to achieve high combustion efficiency. Volume ignition is the use of energy balance relations under the local thermodynamic equilibrium approximation to establish the energy balance equation for thermonuclear systems, and the system ignition threshold is obtained by solving this equation. By understanding the physical process, we believe that the non-equilibrium process is universal to the volume ignition process. Changes in external factors (density, boundaries, albedo, etc.) at the moment of ignition can have a significant impact on the development direction of the system, and an ignition system with a large surface density can nevertheless withstand a large amount of reverse work and continue to burn. The design of the ignition target tries to avoid these factors through margin design, but conversely, the rational use of these laws can further improve the design margin of the capsule. With the aid of big data, the volume ignition method is easier to calculate and has a shorter iteration time. The traditional way is to propose a model, set the material, and then perform the calculation while using big data can set any model and material for calculation. In this paper, a simple comparison will be made to find out that the efficiency of the physical design of a volume ignition target will be effectively improved with the aid of big data. Volume ignition targets can be used not only in Z-pinch-driven systems but also for laser-driven volume ignition. © Published un

关键词： Ignition

来源：评论

学校读者我要写书评

暂无评论

Constraint-Relaxation Approach for Nonnegative Matrix Factorization: A Case Study

Constraint-Relaxation Approach for Nonnegative Matrix Factor...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Jue Wang Naiyang Guan Xuhui Huang Zhigang Luo Science and Technology on Parallel and Distributed Processing Laboratory College of Computer National University of Defense Technology Hunan P.R. China Department of Computer Science and Techonoly College of Computer National University of Defense Technology Hunan P.R. China

ISBN: (纸本)9781479986989

Nonnegative matrix factorization (NMF) is a powerful technique for dimensionality reduction. Conventional NMF algorithms usually keep the matrices W and H nonnegative while iterating. However, to get the NMF of a matrix, it's unnecessary to force the temporary solutions in iterations nonnegative. In this paper, we propose a two-staged approach for NMF. At the relaxation stage, the nonnegative constraint of temporary solutions is relaxed and a real valued matrix factorization is generated. At the constraint stage, the real valued matrix factorization is transformed to a nonnegative matrix factorization by an invertible linear transformation. Based on this approach, we study on exact nonnegative matrix factorization when rank=2. We proved that, given two real valued matrices of rank=2, there exists an invertible linear transformation which can transform the real valued matrices to nonnegative matrices with their product stable. We propose an algorithm to find out the transformation. When rank is higher than 2, this kind of transformation may not exist. In the experiments, it's showed that this approach can reach a nonnegative matrix factorization with lower reconstruction error than conventional methods, and the technique for rank=2 exact NMF works well.

关键词： Algorithm design and analysis Matrix decomposition Additives Computers Transforms Matrix converters Clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Preface

引用

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2003年 2834卷 VI页

作者： Zhou, Xingming National Laboratory for Parallel and Distributed Processing ChangshaHunan China

来源：评论

学校读者我要写书评

暂无评论

DREAMS: Dynamic resource allocation for MapReduce with data skew

DREAMS: Dynamic resource allocation for MapReduce with data ...

引用

IFIP/IEEE International Symposium on Integrated Network Management

作者： Zhihong Liu Qi Zhang Mohamed Faten Zhani Raouf Boutaba Yaping Liu Zhenghu Gong College of Computer National University of Defense Technology Changsha China David R. Cheriton School of Computer Science University of Waterloo Waterloo ON Canada Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha Hunan China

ISBN: (纸本)9781479982424

MapReduce has become a popular model for large-scale data processing in recent years. However, existing MapRe-duce schedulers still suffer from an issue known as partitioning skew, where the output of map tasks is unevenly distributed among reduce tasks. In this paper, we present DREAMS, a framework that provides run-time partitioning skew mitigation. Unlike previous approaches that try to balance the workload of reducers by repartitioning the intermediate data assigned to each reduce task, in DREAMS we cope with partitioning skew by adjusting task run-time resource allocation. We show that our approach allows DREAMS to eliminate the overhead of data repartitioning. Through experiments using both real and synthetic workloads running on a 11-node virtual virtualised Hadoop cluster, we show that DREAMS can effectively mitigate negative impact of partitioning skew, thereby improving job performance by up to 20.3%.

关键词： Resource management Containers Predictive models Mathematical model Monitoring Biomedical monitoring Yarn

来源：评论

学校读者我要写书评

暂无评论

Denoising convolutional autoencoder based B-mode ultrasound tongue image feature extraction

arXiv

引用

arXiv 2019年

作者： Li, Bo Xu, Kele Feng, Dawei Mi, Haibo Wang, Huaimin Zhu, Jian Beijing University of Posts and Telecommunications Automation Dept. Beijing100876 China National Key Lab of Parallel and Distributed Processing Changsha China National University of Defense Technology Changsha China Department of Linguistics University of Michigan Ann Arbor United States

B-mode ultrasound tongue imaging is widely used in the speech production field. However, efficient interpretation is in a great need for the tongue image sequences. Inspired by the recent success of unsupervised deep learning approach, we explore unsupervised convolutional network architecture for the feature extraction in the ultrasound tongue image, which can be helpful for the clinical linguist and phonetics. By quantitative comparison between different unsupervised feature extraction approaches, the denoising convolutional autoencoder (DCAE)-based method outperforms the other feature extraction methods on the reconstruction task and the 2010 silent speech interface challenge. A Word Error Rate of 6.17% is obtained with DCAE, compared to the state-of-the-art value of 6.45% using Discrete cosine transform as the feature extractor. Our codes are available at https://***/DeePBluE666/Source-code1. Copyright © 2019, The Authors. All rights reserved.

关键词： Feature extraction

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：