检索结果-内蒙古大学图书馆

2nd IEEE Asia-Pacific conference on image processing, Electronics and Computers, IPEC 2021

作者： Niu, Simin Taiyuan University of Technology Taiyuan China

ISBN: (纸本)9781728190181

Traditional digital signal processing technology based on DSP and FPGA is more suitable for real-time signal processing, and is limited by data scale and frequency resolution, making it unsuitable for offline data processing, analysis and mining under large-scale data Application. At present, the industrial big data analysis platform can use Spark as a calculation engine for real-time signal processing and offline signal processing acceleration, but the analysis platform lacks mathematical calculation solutions such as digital signal processing suitable for distributed parallel calculation engines. This article is based on time the parity decomposition is selected, and the fast Fourier transform is realized by MATLAB software. Based on an example of the application of the compiled FFT source program, this article analyses the frequency spectrum of discrete-time and continuous-time signals of limited length. © 2021 IEEE.

关键词： MATLAB

来源：评论

学校读者我要写书评

暂无评论

Adaptive Self-Supervised Continual Learning 26

Adaptive Self-Supervised Continual Learning

引用

26th European conference on Artificial Intelligence, ECAI 2023

作者： Wu, Lilei Wang, Zhen Liu, Jie Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha410073 China Laboratory of Digitizing Software for Frontier Equipment National University of Defense Technology Changsha410073 China Tsinghua University Beijing100084 China

ISBN: (纸本)9781643684369

Continual Learning (CL) studies the problem of developing a robust model that can learn new tasks while retaining previously learned knowledge. However, the current CL methods exclusively focus on data with annotations, disregarding that unlabelled data is the mainstream in real-world applications. To close this research gap, this study concentrates on continual self-supervised learning, which is plagued by challenges of memory over-fitting and class imbalance. Besides, these challenges are exacerbated throughout incremental training. Aimed at addressing these challenges from both loss and data perspectives, we introduce a framework, Adaptive Self-supervised Continual Learning (ASCL). Specifically, we devise an Adaptive Sharpness-Aware Minimization (ASAM) module responsible for identifying flatter local minima in the loss landscape with a smaller memory over-fitting risk. Additionally, we design an Adaptive Memory Enhancement (AME) module responsible for rebalancing self-supervised loss with new and old tasks from a data perspective. Finally, the adaptive mechanisms in AME and ASAM modules dynamically adjust the loss landscape sharpness and memory enhancement strength with the feedback of intermediate training results. The results of our extensive experiments demonstrate the state-of-the-art performance of our methods in continual self-supervised learning scenarios across multiple datasets. © 2023 The Authors.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Generic parallel data structures and algorithms to GPU superpixel image segmentation

引用

DISPLAYS 2022年 74卷

作者： Mansouri, Abdelkhalek Creput, Jean-Charles Qiao, Wen-Bao Univ Bourgogne Franche Comte UTBM CIAD UMR 7533 F-90010 Belfort France Beijing Informat Sci & Technol Univ Comp Sch Beijing 100101 Peoples R China

We propose parallel implementation on GPU (graphics processing unit) system for some generic algorithms applied to superpixel image segmentation problem. The aim is to provide standard algorithms based on generic decentralized data structures that could be easily improved and customized on many optimization problems on parallel platforms. Note that superpixel segmentation are clustering algorithms applied to image processing. Two types of algorithms are presented and implemented on GPU based on common parallel data structures. Firstly, we present a parallel implementation of the well-known k-means algorithm with application to 3D data. It is based on a cellular grid subdivision of space that allows closest point findings in constant optimal time for bounded distributions. Secondly, we present an application of the parallel Boruvka minimum spanning forest algorithm to compute watershed segmentation. Both techniques are fully executed on GPU and share the same data structures that embed disjoint-set-trees and distributed-link-lists. We evaluate our k-means approach with regards to state-of-the-art methods, that are, the well known SLIC algorithm, and the adaptive segmentation approach SPASM. We argue that our implementation has the shortest execution time among the tested methods, with a near real time performance and quasi linear acceleration factor, while it provides more regular shape superpixel segmentation based on hexagonal tessellation. Watershed minimum spanning forest method is presented and evaluated accordingly to the same experimental framework.

关键词： k-means Minimum spanning forest Watershed computation parallel computing GPU Superpixel segmentation

来源：评论

学校读者我要写书评

暂无评论

A Dual Network Approach with Enhanced Feature Guidance for SAR-to-Optical image Translation

A Dual Network Approach with Enhanced Feature Guidance for S...

引用

Signal, Information and Data processing (ICSIDP), IEEE International conference on

作者： Jingfei He Yongfei Mao Liangbo Zhao Zihan Cui Yu Wang Hao Shi National Key Laboratory of Science and Technology on Space-Born Intelligent Information Processing Beijing Institute of Technology Beijing China Institute of Remote Sensing Satellite China Academy of Space Technology Beijing China Beijing Institute of Space Mechanics and Electricity China Academy of Space Technology Beijing China

ISBN: (数字)9798331515669

ISBN: (纸本)9798331515676

Synthetic Aperture Radar (SAR) images own dominant merits, such as all-weather and all-day working conditions, but it is difficult for unprofessional people to interpret SAR images. Translating SAR images into optical images to assist interpretation can facilitate the transformation of SAR data into usable information. Therefore, an advanced SAR to Optical (S2O) image translation method utilizing a parallel Generative Adversarial Network (GAN) framework was proposed. This method incorporates an optical image reconstruction network alongside the S2O translation network, enhancing the fidelity and color consistency of the translated images. By employing a domain alignment module and a deep supervision feature loss module, the network effectively utilizes abundant optical image features to overcome the scarcity of SAR-optical image pairs. Experiments conducted on the SEN1-2 dataset demonstrate superior performance over existing methods, particularly in preserving detailed structural information in images.

关键词： Integrated optics Optical losses Translation image color analysis Optical fiber networks Optical imaging Generative adversarial networks Adaptive optics Radar polarimetry Synthetic aperture radar

来源：评论

学校读者我要写书评

暂无评论

Using Stream processing for Real-Time Clock Drift Correction in distributed Data processing Systems

Using Stream Processing for Real-Time Clock Drift Correction...

引用

International Computer Sciences and Information Technologies (CSIT)

作者： Roman Moravskyi Yevheniya Levus Software Department Lviv Polytechnic National University Lviv Ukraine

ISBN: (数字)9798331542634

ISBN: (纸本)9798331542641

In large-scale smart device networks, such as those used in metering systems, the internal clocks of one-way communication devices can drift over time, leading to inaccuracies that impact data-driven decision-making and operational efficiency. Existing solutions often rely on postcollection corrections or synchronization methods that are not feasible for systems requiring real-time accuracy. This study evaluates a method for correcting time drift in smart devices using stream processing. By leveraging a distributed architecture with Apache Kafka, Flink, and OpenSearch, the proposed system processes a continuous stream of device time points at a high rate, ensuring accurate and scalable time correction. Benchmark tests demonstrate the system's capability to maintain low backpressure and high throughput under varying conditions. The findings suggest that the proposed approach is effective for real-time clock drift correction in large-scale networks, offering insights for future integration into Meter Data Management Systems (MDMS).

关键词： Performance evaluation Meters Computer network reliability parallel processing Benchmark testing Data processing Real-time systems Synchronization Smart devices Clocks

来源：评论

学校读者我要写书评

暂无评论

Emotion-Aware Multimodal Pre-training for image-Grounded Emotional Response Generation 27th

Emotion-Aware Multimodal Pre-training for Image-Grounded Emo...

引用

27th International conference on Database Systems for Advanced Applications (DASFAA)

作者： Tian, Zhiliang Wen, Zhihua Wu, Zhenghao Song, Yiping Tang, Jintao Li, Dongsheng Zhang, Nevin L. Hong Kong Univ Sci & Technol Sai Kung Hong Kong Peoples R China Natl Univ Def Technol Sci & Technol Parallel & Distributed Lab Changsha Hunan Peoples R China Natl Univ Def Technol Changsha Hunan Peoples R China

ISBN: (纸本)9783031001291;9783031001284

Face-to-face communication leads to better interactions between speakers than text-to-text conversations since the speakers can capture both textual and visual signals. image-grounded emotional response generation (IgERG) tasks requires chatbots to generate a response with the understanding of both textual contexts and speakers' emotions in visual signals. Pre-training models enhance many NLP and CV tasks and image-text pre-training also helps multimodal tasks. However, existing image-text pre-training methods typically pre-train on images by recognizing or modeling objects, but ignore the emotions expressed in the images. In this paper, we propose several pre-training tasks in a unified framework that not only captures emotions from images but also learns to incorporate the emotion into text generation. The pretraining involves single-modal learning to strengthen the ability to understand images and generate texts. It also involves cross-modal learning to enhance interactions between images and texts. The experiments verify our method in appropriateness, informativeness, and emotion consistency.

关键词： Multimodal Conversation Emotion Pre-training Generation

来源：评论

学校读者我要写书评

暂无评论

Speech Privacy Leakage from Shared Gradients in distributed Learning

Speech Privacy Leakage from Shared Gradients in Distributed ...

引用

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

作者： Zhuohang Li Jiaxin Zhang Jian Liu University of Tennessee Knoxville Intuit AI Research

distributed machine learning paradigms, such as federated learning, have been recently adopted in many privacy-critical applications for speech analysis. However, such frameworks are vulnerable to privacy leakage attacks from shared gradients. Despite extensive efforts in the image domain, the exploration of speech privacy leakage from gradients is quite limited. In this paper, we explore methods for recovering private speech/speaker information from the shared gradients in distributed learning settings. We conduct experiments on a keyword spotting model with two different types of speech features to quantify the amount of leaked information by measuring the similarity between the original and recovered speech signals. We further demonstrate the feasibility of inferring various levels of side-channel information, including speech content and speaker identity, under the distributed learning framework without accessing the user’s data.

关键词： Computer aided instruction Privacy Data privacy Speech analysis Distance learning Vocoders distributed databases

来源：评论

学校读者我要写书评

暂无评论

Multi-channel Walk Embedding Based Ethereum Phishing Scam Detection Method

Multi-channel Walk Embedding Based Ethereum Phishing Scam De...

引用

International conference on parallel and distributed Systems (ICPADS)

作者： Jinghao Li Hexiao Li Na Cheng Wuqing Zhang Ya Xu Hengyang Zhang Jianbin Li School of Control and Computer Engineeging North China Electric Power University Beijing China

With the prevalent adoption of blockchain in the financial system, there has been an increase in phishing scams on cryptocurrency platforms such as Ethereum, and an effective anomaly detection method is urgently required. The latest studies have focused on anomaly identification using natural language processing techniques or constructing simple static graphs. However, the existing methods are insufficient to convey the diversity of connectivity patterns in the Ethereum transaction network concerning amount and time. To this end, we proposed a novel transaction network embedding algorithm transE based on the multi-channel random walk to model the detection of Ethereum phishing scam accounts as a multigraph node classification task. Specifically, we first model the Ethereum transaction as a time-amount directed multigraph. Then, the hybrid feature representation of network nodes is learned via transE from their local and global neighbours, which uses the attention mechanism to maximize the probability of preserving node network neighbours. Ultimately, we employ visualization techniques and machine learning models to validate the effectiveness of the algorithms, and the model with the top performance is picked for Ethereum account classification. Experimental results indicate that the embedding vector extracted by transE improves the detection accuracy of Ethereum phishing accounts in the different classification tasks.

关键词： Visualization Machine learning algorithms Phishing Computational modeling Machine learning Feature extraction Natural language processing

来源：评论

学校读者我要写书评

暂无评论

Graphics processing methods Based on Deep Learning in the Context of Big Data

Graphics Processing Methods Based on Deep Learning in the Co...

引用

International conference on Multi-modal Information Analytics, MMIA 2021

作者： Yang, Wei College of Information Baoshan University BaoshanYunnan678000 China

ISBN: (纸本)9783030748135

The purpose of this article is to conduct research on graphics processing (GP) methods based on DL in the context of big data (BD). First, this article studies the characteristics of current BD and investigates related applications. Then based on DL, this article proposes a method of processing graphics content. This method uses three different DL network structures to extract graphic features. In this paper, the OpenMP parallel model is used to optimize the graph search algorithm in parallel on the CPU platform, and the algorithm is optimized in parallel by using the principle of program locality, reducing synchronization overhead and load balancing. Thirdly, in view of the irregularity of the memory access of the graph search algorithm, the FPGA algorithm hardware accelerator is customized using BD technology. Finally, it introduces the development tools of DL algorithm hardware acceleration, and compares them with traditional processing methods. Comparative experiments show that the image processing method based on DL proposed in this paper has an average recognition rate of about 10% higher than traditional processing methods, and has better results, providing an important reference for the development of image processing. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Big data

来源：评论

学校读者我要写书评

暂无评论

Towards Network-accelerated ML-based distributed Computer Vision Systems 27

Towards Network-accelerated ML-based Distributed Computer Vi...

引用

27th IEEE International conference on parallel and distributed Systems, ICPADS 2021

作者： Siddique, Hisham Neves, Miguel Kuzniar, Carson Haque, Israat Dalhousie University Faculty of Computer Science HalifaxNS Canada

ISBN: (纸本)9781665408783

Computer vision is a crucial component in many modern applications (e.g., medical image analysis, environmental monitoring and self-driving cars). However, their stringent computational, latency and bandwidth requirements still pose a huge challenge to system architects, which must seek for alternatives to both the limited resources (e.g., low-end CPU) on client devices and the hurdles of moving data from clients to cloud/edge servers for analysis. In this work, we advocate for the usage of emerging programmable network devices to speed up ML-based computer vision tasks, particularly image classification, on resource constrained environments. To take the first step towards this new paradigm, we propose NetPixel, a framework that enables P4-programmable switches to classify images in realtime, accurately and at scale. We implemented a prototype of NetPixel in a software switch to show its feasibility and conducted a preliminary evaluation on widely adopted datasets. Our results show that NetPixel can classify images with an accuracy within 8% that of a server-based implementation even for shallow classifiers and low-resolution images. © 2021 IEEE.

关键词： Network protocols

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：