检索结果-内蒙古大学图书馆

International conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Mingyu Liu Yijie Wang Xiaohui Zhou Yongjun Wang National Key Laboratory of Parallel and Distributed Computing College of Computer Science and Technology National University of Defense Technology Changsha China College of Computer Science and Technology National University of Defense Technology Changsha China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Multivariate time series anomaly detection (MTAD) poses a challenge due to temporal and feature dependencies. The critical aspects of enhancing the detection performance lie in accurately capturing the dependencies between variables within the sliding window and effectively leveraging them. Existing studies rely on domain knowledge to pre-set the window size, and overlook the strength of dependencies while calculating direction based on variable similarity. This paper proposes GSLTE, a graph structure learning method for MTAD. GSLTE employs Fast Fourier Transform to conduct iterative segmentation of the whole series, selecting the dominant Fourier frequency as the window size for each subsequence within the minimum interval. GSLTE quantifies the direction and strength of the dependencies based on variable-lag transfer entropy which is achieved through Dynamic Time Warping method to learn asymmetric links between variables. Extensive experiments show that GNN-based MTAD methods applying GSLTE can further improve anomaly detection performance while outperforming state-of-the-art competitors.

关键词： Learning systems Time-frequency analysis Fast Fourier transforms Time series analysis Signal processing Feature extraction Entropy Iterative methods Speech processing Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

TMan: A High-Performance Trajectory Data Management System Based on Key-Value Stores

TMan: A High-Performance Trajectory Data Management System B...

引用

International conference on Data Engineering

作者： Huajun He Zihang Xu Ruiyuan Li Jie Bao Tianrui Li Yu Zheng School of Computing and Artificial Intelligence Southwest Jiaotong University China JD iCity JD Technology Beijing China JD Intelligent Cities Research China Chongqing University China

ISBN: (数字)9798350317152

ISBN: (纸本)9798350317169

The effective management of trajectory data heavily relies on the utilization of fundamental spatio-temporal queries. The surge in trajectory data, with its dynamic spatio-temporal properties, poses notable management challenges. Existing systems are inadequate in providing fine-grained trajectory representations and efficient architecture for processing queries, leading to significant computational overhead. This paper introduces TMan to address these challenges. First, TMan presents two innovative index structures that precisely capture the spatio-temporal characteristics of trajectory data. Compared to the state-of-the-art indexes, our indexes for temporal range and spatial range queries can reduce the number of retrievals by up to 77% and 83%, respectively. Next, TMan devises concise and effective encoding methods for these indexes. Leveraging these indexes, TMan provides a distributed storage structure and an index caching mechanism for efficiently managing trajectories in key-value data stores. Moreover, TMan introduces a parallel query processing approach incorporating a push-down strategy to enhance the efficiency of fundamental queries. Extensive experimental results demonstrate that TMan's index structures and architecture outperform the baselines.

关键词： Shape Query processing distributed databases Computer architecture Data engineering Encoding Trajectory

来源：评论

学校读者我要写书评

暂无评论

Annotations Are Not All You Need: A Cross-modal Knowledge Transfer Network for Unsupervised Temporal Sentence Grounding

Annotations Are Not All You Need: A Cross-modal Knowledge Tr...

引用

conference on Empirical methods in Natural Language processing (EMNLP)

作者： Fang, Xiang Liu, Daizong Fang, Wanlong Zhou, Pan Cheng, Yu Tang, Keke Zou, Kai Huazhong Univ Sci & Technol Hubei Engn Res Ctr Big Data Secur Sch Cyber Sci & Engn Hubei Key Lab Distributed Syst Secur Wuhan Peoples R China Peking Univ Beijing Peoples R China Henan Univ Kaifeng Peoples R China Chinese Univ Hong Kong Hong Kong Peoples R China Guangzhou Univ Guangzhou Peoples R China Protagolabs Inc Vienna VA USA

ISBN: (纸本)9798891760615

This paper addresses the task of temporal sentence grounding (TSG). Although many respectable works have made decent achievements in this important topic, they severely rely on massive expensive video-query paired annotations, which require a tremendous amount of human effort to collect in real-world applications. To this end, in this paper, we target a more practical but challenging TSG setting: unsupervised temporal sentence grounding, where both paired video-query and segment boundary annotations are unavailable during the network training. Considering that some other cross-modal tasks provide many easily available yet cheap labels, we tend to collect and transfer their simple cross-modal alignment knowledge into our complex scenarios: 1) We first explore the entity-aware objectguided appearance knowledge from the paired image-Noun task, and adapt them into each independent video frame;2) Then, we extract the event-aware action representation from the paired Video-Verb task, and further refine the action representation into more practical but complicated real-world cases by a newly proposed copy-paste approach;3) By modulating and transferring both appearance and action knowledge into our challenging unsupervised task, our model can directly utilize this general knowledge to correlate videos and queries, and accurately retrieve the relevant segment without training. Extensive experiments on two challenging datasets (ActivityNet Captions and Charades-STA) show our effectiveness, outperforming existing unsupervised methods and even competitively beating supervised works.

关键词： Knowledge management

来源：评论

学校读者我要写书评

暂无评论

High-Resolution Projection Network combining High-Resolution Optical Flow Compensation for Video Super-Resolution

High-Resolution Projection Network combining High-Resolution...

引用

image processing and Media Computing (ICIPMC), International conference on

作者： Yifei Sun Zhengxia Chen Yuying Jin Xiaoyi Feng School of Electronics and Information Northwestern Polytechnical University Xi'an China Planning Department Xi'an Aerospace Propulsion Institute Xi'an China College of Information Science and Technology Northwest University Xi'an China

ISBN: (数字)9798350386660

ISBN: (纸本)9798350386677

The Super-Resolution (SR) task is to generate high-resolution (HR) images/videos using low-resolution (LR) ones. At present, the Single image Super-Resolution (SISR) methods have achieved superior performance. However, Video Super-Resolution (VSR) methods still have some disadvantages such as edge blurring, high-frequency details missing and motion artifacts. This paper proposes an end-to-end HR feature projection VSR network HOF-HRPN, which consists of a High-Resolution Optical Flow (HOF) and a High-Resolution Projection Network (HRPN). The HOF network estimates the HR optical flow and compensates it to the neighboring frames to achieve accurate frame alignment. The HRPN is composed of a multi frame feature projection channel and a single frame SR channel in parallel. HRPN takes into account the advantages that single frame SR and VSR can extract missing high-frequency details from intra-frame and inter-frame, respectively. It fuses high-frequency details obtained from multi-scale LR projection learning and single frame SR results. A large number of comparative experiments based on public datasets verify that HOF-HRPN is robust and can recover accurate pixel values, clear edges and rich textures.

关键词： Accuracy image edge detection Superresolution Streaming media Media Robustness Real-time systems

来源：评论

学校读者我要写书评

暂无评论

ByteZip: Efficient Lossless Compression for Structured Byte Streams Using DNNs

ByteZip: Efficient Lossless Compression for Structured Byte ...

引用

International Joint conference on Neural Networks (IJCNN)

作者： Parvathy Ramakrishnan P Satyajit Das Department of Data Science Indian Institute of Technology Palakkad Kerala India

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

Data compression plays a key role in efficient data storage, transmission, and processing. With the fast development of deep learning techniques, deep neural networks have been used in this field to achieve a higher compression rate. Deep learning-based general-purpose lossless compression techniques are formulated as an autoregressive sequential prediction problem. These methods are state-of-the-art in terms of compression ratio but not practical due to runtime and resource constraints. Recent advances in lossless image compression using non-autoregressive methods for probability modeling prove to be a faster and more practical approach. In this paper, we propose ByteZip, a lossless compression method based on the non-autoregressive approach for known or defined structured byte streams. ByteZip involves hierarchical probabilistic modeling using autoencoders and density mixture models. This approach reduces the overhead of sequential processing. The goal is to design a practical lossless compressor with faster compression and decompression along with a competitive compression ratio. Experiments show that the proposed approach achieves a 64× higher compression speed than the state-of-the-art transformer-based model TRACE with an overhead of only 5% less size reduction on average. Our approach outperforms general-purpose compressors such as Gzip (23% more size reduction on average) and 7z (16% more size reduction on average).

关键词： image coding Runtime Pipelines Data compression Predictive models parallel processing Transformers

来源：评论

学校读者我要写书评

暂无评论

Optimizing Privacy While Limiting Information Loss in distributed Data Anonymization

Optimizing Privacy While Limiting Information Loss in Distri...

引用

IEEE International conference on Big Data

作者： Arseme Vadèle Djeufack Nanfack Gaël Le Mahec Gilles Dequen MIS Laboratory University of Picardie Jules Verne Amiens France

ISBN: (数字)9798350362480

ISBN: (纸本)9798350362497

k-anonymity [1], [2], aims to ensure that individual data cannot be distinguished from that of at least (k−1) others in the same database, regardless of additional knowledge. However, this process involves modifying data, resulting in information loss (IL). Finding the optimal solution to minimize this loss is NP-Hard [3], [4], prompting the development of heuristics. While some of these solutions are quadratic, they are impractical for large databases, leading to proposals for distributed memory environments. In these environments, some methods utilize horizontal partitioning of the database to speed up execution time, with each processor handling multiple records simultaneously. However, as the number of processors increases, the size of subsets handled by each processor decreases, leading to increased information loss, especially as k increases. This paper addresses the problem of minimizing information loss when anonymizing large databases. We propose an approach that exploits parallelism for cluster-computing using horizontal partitioning with overlaps (i.e., where partitions have common rows) to enhance information loss for the anonymization of databases. After anonymization, anonymous subsets are aggregated to create a global anonymized database by removing duplicate records. Our proposed algorithm employs parallel hierarchical aggregation that chooses a better version of an anonymized record among the different versions where it overlaps. Experimental results show that our approach is approximately 80× faster and incurs less information loss than the centralized GkAA [5], [6] algorithm.

关键词： Data privacy distributed databases Big Data parallel processing Approximation algorithms Minimization Information filtering Partitioning algorithms Proposals Information integrity

来源：评论

学校读者我要写书评

暂无评论

HMA: An Efficient Training Method for NLP Models 2021

HMA: An Efficient Training Method for NLP Models

引用

5th International conference on Innovation in Artificial Intelligence (ICIAI)

作者： Yang, Yuetong Lai, Zhiquan Cai, Lei Li, Dongsheng Natl Univ Def Technol Sch Comp Natl Key Lab Parallel & Distributed Proc Changsha Hunan Peoples R China

ISBN: (纸本)9781450388634

Nowadays, deep learning has been widely used for solving natural language processing (NLP) problems. Embedding matrices are common-used in the NLP deep models for automatical feature learning. However, the sparsity of embedding matrices makes it challenging to efficiently train the NLP models in data parallelism. When training with synchronous optimization methods, the aggregation on sparse gradients brings high communication cost and low scalability for distributed training. In this paper, we combine Model Average (MA) and synchronous optimization methods together, and propose HMA, a hybrid training method for NLP deep models. Furthermore, we implement HMA method in Horovod+TensorFlow training framework and conduct experimental evaluation with representative NLP models. For NLP models with a large number of sparse parameters, HMA saves over 30% wall-clock time compared with the state-of-the-art distributed training framework, while maintaining the same final training loss.

关键词： distributed training deep learning model average local sgd

来源：评论

学校读者我要写书评

暂无评论

parallel Optimisation and Implementation of a Real-Time Back Projection (BP) Algorithm for SAR Based on FPGA

引用

SENSORS 2022年第6期22卷 2292页

作者： Cao, Yue Guo, Shuchen Jiang, Shuai Zhou, Xuan Wang, Xiaobei Luo, Yunhua Yu, Zhongjun Zhang, Zhimin Deng, Yunkai Chinese Acad Sci Aerosp Informat Res Inst Beijing 100094 Peoples R China Univ Chinese Acad Sci Sch Elect Elect & Commun Engn Beijing 101408 Peoples R China

This study conducts an in-depth evaluation of imaging algorithms and software and hardware architectures to meet the capability requirements of real-time image acquisition systems, such as spaceborne and airborne synthetic aperture radar (SAR) systems. By analysing the principles and models of SAR imaging, this research creatively puts forward the fully parallel processing architecture for the back projection (BP) algorithm based on Field-Programmable Gate Array (FPGA). The processing time consumption has significant advantages compared with existing methods. This article describes the BP imaging algorithm, which stands out with its high processing accuracy and two-dimensional decoupling of distance and azimuth, and analyses the algorithmic flow, operation, and storage requirements. The algorithm is divided into five core operations: range pulse compression, upsampling, oblique distance calculation, data reading, and phase accumulation. The architecture and optimisation of the algorithm are presented, and the optimisation methods are described in detail from the perspective of algorithm flow, fixed-point operation, parallel processing, and distributed storage. Next, the maximum resource utilisation rate of the hardware platform in this study is found to be more than 80%, the system power consumption is 21.073 W, and the processing time efficiency is better than designs with other FPGA, DSP, GPU, and CPU. Finally, the correctness of the processing results is verified using actual data. The experimental results showed that 1.1 s were required to generate an image with a size of 900 x 900 pixels at a 200 MHz clock rate. This technology can solve the multi-mode, multi-resolution, and multi-geometry signal processing problems in an integrated manner, thus laying a foundation for the development of a new, high-performance, SAR system for real-time imaging processing.

关键词： synthetic aperture radar (SAR) back-projection algorithm (BP) real-time image processing field-programmable gate array (FPGA)

来源：评论

学校读者我要写书评

暂无评论

DFBuffer: High-performance data forwarding software optimized for single-process I/O scenarios

DFBuffer: High-performance data forwarding software optimize...

引用

International conference on parallel and distributed Systems (ICPADS)

作者： Xiaobin He Wei Xiao Xun Deng Qi Chen Bin Yang Zuoning Chen National Research Center of Parallel Computer Engineering & Technology Beijing China Wuhan National Laboratory for Optoelectronics Huazhong University of Science and Technology Wuhan China Department of Computer Science and Technology Tsinghua University Beijing China Chinese Academy of Engineering Beijing China

Most supercomputers adopt a data forwarding architecture to achieve storage scalability. However, it results in a significant reduction in single-process bandwidth compared to direct file system access. Moreover, considering that a majority of applications uses only a single process for writing and reading data, the low single-process performance also leads to a time overhead for these applications. This paper proposes an userspace forwarding mechanism DFBUFFER with two performance optimization methods: user-space multi-thread request processing and data write buffer in a unit of file. The client of DFBUFFER is embedded in the application as a library reducing the software overhead, and the server implements multi-thread I/O request processing to improve bandwidth efficiency. The data write buffer can asynchronously handle write requests, which accelerates the write bandwidth of compute nodes. We evaluate DFBUFFER on the Sunway exascale prototype system. The results indicate that in the regular mode of DFBUFFER, both the write and read latency are reduced, and the write bandwidth and large-block read bandwidth of single-process are increased by 1.8 times and 2.8 times respectively. The DFBUFFER buffer mode increases the write bandwidth of a single process by 0.8 times over the regular mode. Although the performance advantage of the regular mode of DFBUFFER gradually weakens with the increase of concurrent processes, the DFBUFFER buffer mode has the effect of improving the write bandwidth, the 64-IO-processes application is increased by 0.2 times.

关键词： Spectral efficiency Scalability Prototypes Bandwidth Computer architecture Writing Software

来源：评论

学校读者我要写书评

暂无评论

Path planning and image classification algorithms for glass cleaning robots based on hybrid path planning and convolutional neural networks

Path planning and image classification algorithms for glass ...

引用

Computers, Information processing and Advanced Education (CIPAE), 2020 International conference on

作者： Zequn Li Xiao Han Xiao Wang Yu Li Xingyu Zhu Hohai University Nanjing China

With the development of the times and the progress of society, the number of high-rise buildings is increasing, and the cleaning of the outer glass at height is extremely difficult, relying on manual cleaning with high safety risks and low efficiency. The trend is to use automated control products such as cleaning robots to take on some of the difficult cleaning work. Path planning and image recognition for glass cleaning robots is a hot issue for cleaning robots to achieve efficient cleaning. In this paper, a path planning and image recognition method for glass cleaning robots based on hybrid path planning and convolutional neural networks is proposed to address this hot problem. The cleaning robot adopts a hybrid path of contour-parallel path and method-parallel path to achieve full-coverage path planning, and uses a portable camera to take photos of the glass, and uses a convolutional neural network model to determine the degree of dirty glass, and uses this as the basis for different cleaning processes, effectively improving the efficiency of cleaning and the utilisation of resources. In this paper, we design an intelligent control algorithm for a glass cleaning robot that combines artificial intelligence and 3D printing path planning generation by combining two of these methods, which helps to improve the cleaning effect of the cleaning robot.

关键词：

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：