检索结果-内蒙古大学图书馆

Magnifier: A Chiplet Feature-Aware Test Case Generation Method for Deep Learning Accelerators

IEEE Transactions on computer-Aided Design of Integrated Circuits and Systems 2025年

作者： Li, Boyu Zhu, Zongwei Liu, Weihong Cao, Qianyue Li, Changlong Ji, Cheng Li, Xi Zhou, Xuehai University of Science and Technology of China School of Computer Science and Technology Hefei230026 China University of Science and Technology of China Suzhou Institute for Advanced Research Suzhou215123 China University of Science and Technology of China School of Software Engineering Hefei230026 China University of Science and Technology of China China and Suzhou Institute for Advanced Research Suzhou215123 China East China Normal University School of Computer Science and Technology Shanghai200063 China Nanjing University of Science and Technology School of Computer Science and Engineering Nanjing210094 China

The development of deep learning has led to increasing demands for computation and memory, making multi-chiplet accelerators a powerful solution. Multi-chiplet accelerators require more precise consideration of hardware configurations and mapping schemes in terms of computation, memory, and communication patterns compared to monolithic designs, in order to avoid underutilization of performance. However, there is currently a lack of performance testing methods specifically tailored for multi-chiplet accelerators. Existing testing methods primarily focus on correctness testing and do not address potential performance issues from a hardware perspective. To address these issues, this paper proposes Magnifier: A test case generation method for performance testing of multi-chiplet accelerators. Firstly, we analyze typical multi-chiplet accelerator prototype from the perspectives of computation, memory, and communication patterns, and summarize a chiplet feature-Aware operator task set. Next, we define the test evaluation metric IPPstd and use a candidate operator set to construct a sampling space for model-level test cases. Finally, we build a GAN to learn the distribution of high-diversity test cases, enabling the rapid generation of high-quality test cases. We validate the proposed method on both simulated and real multi-chiplet accelerators. Experiments show that Magnifier can improve the metric of test cases by up to 3.42 times and significantly reduce generation time, providing valuable insights for optimizing the hardware and software of multi-chiplet accelerators. © 1982-2012 IEEE.

关键词： Acceleration

来源：评论

学校读者我要写书评

暂无评论

A Systemic Big Data Framework for the Charging Pile Business 22

A Systemic Big Data Framework for the Charging Pile Business

引用

Proceedings of the 7th International Conference on Information Systems engineering

作者： Licheng Xu Shuai Xie Department of Software Engineering Universiti Malaya Malaysia Faculty of Computer Science and Information Technology Universiti Malaya Malaysia

ISBN: (纸本)9781450397889

There are some problems in the electric vehicle (EV) charging pile industry, such as the unreasonable location of charging station construction, low utilization rate of charging piles, and imprecise marketing strategies, which may elicit a negative response from EV users, cause colossal waste of resources and hinder the development of EVs. Based on the relevant charging pile data to be analyzed, this paper combines big data-related technologies to propose a big data framework for analyzing the charging pile data to solve the common charging service problems from a systemic perspective. The purpose of the framework is to provide decision-making reference information for charging pile operators, charging pile application operators and charging service marketers to improve the charging pile business more effectively. The framework is demonstrated by functional structure and technical structure.

关键词： Charging pile data Big Data framework Decision-making reference Charging business improvement

来源：评论

学校读者我要写书评

暂无评论

Self-Bidirectional Decoupled Distillation for Time Series Classification

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第8期5卷 4101-4110页

作者： Xiao, Zhiwen Xing, Huanlai Qu, Rong Li, Hui Feng, Li Zhao, Bowen Yang, Jiayi Southwest Jiaotong University School of Computing and Artificial Intelligence Chengdu610031 China Tangshan Institute of Southwest Jiaotong University Tangshan063000 China Engineering Research Center of Sustainable Urban Intelligent Transportation Ministry of Education Chengdu611756 China University of Nottingham School of Computer Science NottinghamNG7 2RD 455356 United Kingdom Xi'an Jiaotong University School of Mathematics and Statistics Xi'an710049 China Tongji University School of Software Engineering Shanghai201804 China

Over the years, many deep learning algorithms have been developed for time series classification (TSC). A learning model's performance usually depends on the quality of the semantic information extracted from lower and higher levels within the representation hierarchy. Efficiently promoting mutual learning between higher and lower levels is vital to enhance the model's performance during model learning. To this end, we propose a self-bidirectional decoupled distillation (self-BiDecKD) method for TSC. Unlike most self-distillation algorithms that usually transfer the target-class knowledge from higher to lower levels, self-BiDecKD encourages the output of the output layer and the output of each lower level block to form a bidirectional decoupled knowledge distillation (KD) pair. The bidirectional decoupled KD promotes mutual learning between lower and higher level semantic information and extracts the knowledge hidden in the target and nontarget classes, helping self-BiDecKD capture rich representations from the data. Experimental results show that compared with a number of self-distillation algorithms, self-BiDecKD wins 35 out of 85 University of California, Riverside (UCR) 2018 datasets and achieves the smallest AVeraGe (AVG)_rank score, namely 3.2882. In particular, compared with a nonself-distillation baseline, self-BiDecKD results in 58/8/19 regarding 'win'/'tie'/'lose.' © 2020 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Transformers in Small Object Detection: A Benchmark and Survey of State-of-the-Art

arXiv

引用

arXiv 2023年

作者： Rekavandi, Aref Miri Rashidi, Shima Boussaid, Farid Hoefs, Stephen Akbas, Emre Bennamoun, Mohammed Department of Computer Science and Software Engineering The University of Western Australia Australia Department of Electrical Electronics and Computer Engineering The University of Western Australia Australia Defence Science and Technology Group Australia Department of Computer Engineering Middle East Technical University Turkey

Transformers have rapidly gained popularity in computer vision, especially in the field of object recognition and detection. Upon examining the outcomes of state-of-the-art object detection methods, we noticed that transformers consistently outperformed well-established CNN-based detectors in almost every video or image dataset. While transformer-based approaches remain at the forefront of small object detection (SOD) techniques, this paper aims to explore the performance benefits offered by such extensive networks and identify potential reasons for their SOD superiority. Small objects have been identified as one of the most challenging object types in detection frameworks due to their low visibility. We aim to investigate potential strategies that could enhance transformers’ performance in SOD. This survey presents a taxonomy of over 60 research studies on developed transformers for the task of SOD, spanning the years 2020 to 2023. These studies encompass a variety of detection applications, including small object detection in generic images, aerial images, medical images, active millimeter images, underwater images, and videos. We also compile and present a list of 12 large-scale datasets suitable for SOD that were overlooked in previous studies and compare the performance of the reviewed studies using popular metrics such as mean Average Precision (mAP), Frames Per Second (FPS), number of parameters, and more. Researchers can keep track of newer studies on our web page, which is available at: https://***/arekavandi/Transformer-SOD. © 2023, CC BY-NC-SA.

关键词： Object recognition

来源：评论

学校读者我要写书评

暂无评论

Systematically Generated Formulas for Spectrum-Based Fault Localization

Systematically Generated Formulas for Spectrum-Based Fault L...

引用

IEEE International Conference on software Testing Verification and Validation Workshop, ICSTW

作者： Qusay Idrees Sarhan Tamás Gergely Árpád Beszédes Department of Software Engineering University of Szeged Szeged Hungary Department of Computer Science University of Duhok Duhok Iraq

The basic element in Spectrum-Based Fault Localization (SBFL) are the risk evaluation formulas, which calculate a suspiciousness score for each program element based on test coverage and test case outcome information. This score can be used in debugging to identify the faulty element more efficiently. A large number of manually crafted formulas have been proposed, but a line of research tries to generate formulas (semi-)automatically. Some of these approaches are based on heuristic search (e.g., genetic algorithms), and researchers started only recently examining systematic ways to generate all possible formulas corresponding to a particular class of formula structures. In a recent work, we explored a very simple formula template as a proof of concept but this research failed to find a new formula that outperformed already published ones. In this paper, we take a next step and investigate a class of formula templates that are more elaborate but still feasible to explore fully for generating new formulas. Many of the generated formulas cover some well-known existing ones, but we were also able to find two new ones that are superior to the majority of the previously published formulas (evaluated on the Defects4J dataset) and are not present in literature.

关键词： Location awareness Fault diagnosis software testing Systematics Conferences Debugging Benchmark testing

来源：评论

学校读者我要写书评

暂无评论

Multiobjective Band Selection Approach Via an Adaptive Particle Swarm Optimizer for Remote Sensing Hyperspectral Images

SSRN

引用

SSRN 2023年

作者： Zhang, Yuze Lin, Qiuzhen Li, Lingjie Xiao, Zhijiao Ming, Zhong Leung, Victor C.M.L. College of Computer Science and Software Engineering Shenzhen University Shenzhen China Shenzhen China Department of Electrical and Computer Engineering The University of British Columbia VancouverBC Canada

Band selection is an important technique in remote sensing hyperspectral image (HSI) that captures informative bands and removes redundant bands, thereby reducing the computational overhead of model and improving classification performance. Recently, evolutionary algorithm with good search capability has been successfully extended as efficient BS methods for HSIs. However, most of existing EA-based BS methods still face following challenges: 1) Most of them are designed based on single search strategy, which are insufficient to find promising bands subset and tend to fall into the local optimum;2) Most of them ignore the problem of potential duplicate bands existing in the solution vectors, which leads to the performance degradation. To address these issues, this paper proposes an effective unsupervised multiobjective BS approach via an adaptive particle swarm optimizer. First, an unsupervised multiobjective BS model is designed, which applies information entropy to measure the amount of information contained in bands and structural similarity measure to evaluate the correlation among bands. Second, to effectively solve the BS model constructed above, an adaptive particle swarm optimizer is proposed, which first applies a pre-scoring mechanism to estimate the evolutionary states (i.e., diversity and convergence degree) of particles, and then adaptively selects different PSO search strategies according to their corresponding evolutionary states. In addition, a self-repair mechanism is designed to correct the violating individuals with duplicate bands. Finally, to evaluate the performance of APSO-BS, four common HSI datasets are adopted in our experiments, and extensive results show that our proposed method outperforms several state-of-the-art BS methods. © 2023, The Authors. All rights reserved.

关键词： Multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

PCHM-Net: A New Point Cloud Compression Framework for Both Human Vision and Machine Vision

PCHM-Net: A New Point Cloud Compression Framework for Both H...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Lei Liu Zhihao Hu Jing Zhang School of Computer Science and Engineering Beihang University Beijing China School of Software Beihang University Beijing China

Recently, point cloud data has attracted increasing attention in various machine vision tasks like classification and detection. However, directly transmitting the raw point cloud for such machine vision tasks will bring a huge bit-rate cost. In this work, we propose a new point cloud compression framework called PCHM-Net for both human vision and machine vision. Our proposed PCHM-Net adopts a two-branch structure with the shared octree-based compression module. To better compress the point cloud data and save bit-rate for machine vision tasks, we use the point cloud selection module to select a sparse set of points before octree construction, which allows us to use deeper octree structure and thus better reconstruct the point cloud coordinates for more discriminative feature extraction. We further propose a global feature aggregation-based classification module to deal with the sparse point cloud classification task. Comprehensive experiments on various point cloud benchmark datasets (e.g., ModelNet, ShapeNet and ScanNet) demonstrate that our newly proposed PCHM-Net achieves promising coding performance for both human vision and machine vision.

关键词：

来源：评论

学校读者我要写书评

暂无评论

EBFT: Efficient BFT Consensus through Shortcut Replies 21

EBFT: Efficient BFT Consensus through Shortcut Replies

引用

21st IEEE International Symposium on Parallel and Distributed Processing with Applications, 13th IEEE International Conference on Big Data and Cloud Computing, 16th IEEE International Conference on Social Computing and Networking and 13th International Conference on Sustainable Computing and Communications, ISPA/BDCloud/SocialCom/SustainCom 2023

作者： Lei, Qingsheng Xia, Yongjun Guo, Qitao Hao, Rui Dai, Xiaohai Dai, Weiqi Economic & Technology Research Institute State Grid Hubei Electric Power Company Limited China Hubei Central China Technology Development of Electric Power Company Limited China Software Institute Nanjing University China Huazhong University of Science and Technology School of Computer Science and Technology China Huazhong University of Science and Technology School of Cyber Science and Engineering China

ISBN: (纸本)9798350329223

As the core building block of blockchain technology, Byzantine Fault Tolerance (BFT) consensus has gathered renewed attention during the past decade. A large number of BFT consensus protocols have been proposed since its inception in the 1980s, the best-known of which is Practical Byzantine Fault Tolerance (PBFT). Generally speaking, PBFT consists of two sub-protocols: normal-case protocol and view-change protocol. However, it requires 7 δ for a client to commit a request in the normal case, where δ represents the actual network delay. If a request happens to go through the view change process, it will take additional 7 δ for a client to commit a request. To reduce the consensus latency, we propose a novel BFT consensus protocol named EBFT, which introduces the shortcut replies to the consensus design. With these shortcut replies, EBFT reduces the normal-case latency and view-change latency to 5 δ and 2 δ, respectively. Extensive experiments are conducted to evaluate EBFT, whose results demonstrate its effectiveness. © 2023 IEEE.

关键词： Fault tolerance

来源：评论

学校读者我要写书评

暂无评论

Fine-Tuning Vision Transformer for Arabic Sign Language Video Recognition on Augmented Small-Scale Dataset

Fine-Tuning Vision Transformer for Arabic Sign Language Vide...

引用

2023 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2023

作者： Gochoo, Munkhjargal Batnasan, Ganzorig Ahmed, Ahmed Abdelhadi Otgonbold, Munkh-Erdene Alnajjar, Fady Shih, Timothy K. Tan, Tan-Hsu Wee, Lai Khin United Arab Emirates University Dept. of Computer Science and Software Engineering Al Ain United Arab Emirates College of EECS National Central University Taoyuan Taiwan National Taipei University of Technology Dept. of Electrical Engineering Taipei Taiwan University of Malaya Dept. of Biomedical Engineering Kuala Lumpur Malaysia

ISBN: (纸本)9798350337020

With the rise of AI, the recognition of Sign Language (SL) through sign-to-text has gained significance in the field of computer vision and deep machine learning. However, there are only a few medium to large open datasets available for this task, as it requires a vast dataset of thousands of signs for words/phrases in different environments, which is a time-consuming and tedious process. Furthermore, there has been very little effort towards Arabic Sign Language Recognition (ArSLR). This research paper presents the results of fine-tuning the Vision Transformer (ViT) model on a small-scale in-house dataset of ArSL. The main goal is to attain satisfactory results by utilizing minimal computing power and a small dataset involving less than 10 individuals, with only one recording made for each sign in every environment. The dataset comprises 49 classes/signs, all of which were made with two hands and belong to the Level I category in terms of popularity. To enhance the dataset, three types of augmentations - translation, shear, and rotation were employed. The ViT model, pre-trained on the Kinetics dataset, was trained on the variation of augmented datasets with 2 to 40 times samples for each original video, where the training set includes original and augmented videos of 8 volunteers and the test set includes only original videos of one particular volunteer. Experimental results reveal that the combination of rotation and shear outperformed the others, achieving an accuracy of 93% on the 20 times augmented samples per class per signer dataset. We believe this study sheds light on small-scale dataset-based SLR tasks and video/action recognition in general. © 2023 IEEE.

关键词： Arabic Sign Language Augmentation Deep Learning Smale-scale dataset Vision Transformer ViT

来源：评论

学校读者我要写书评

暂无评论

Public Health Emergencies: Health Data and Public Health Surveillance

Public Health Emergencies: Health Data and Public Health Sur...

引用

Digital Health (ICDH), IEEE International Conference on

作者： Mohammad Nazayer Samaneh Madanian Hamidreza Rasouli Panah Dave Parry Department of Computer Science and Software Engineering AUT Auckland New Zealand School of IT Murdoch University Perth Australia

After COVID-19, the healthcare system’s ineffectiveness in managing pandemics or public health emergencies is significantly highlighted. Based on the increase in the frequency of pandemics, our objective in this research is to define and propose an integrated health system to support healthcare preparedness for future public health emergencies. This system can support managers and authorities in healthcare and disaster management, through data collection, sharing, and analysis, which would ultimately enhance the effectiveness of managing an outbreak before becoming a pandemic.

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：