检索结果-内蒙古大学图书馆

A Primal-Dual SGD Algorithm for Distributed Nonconvex Optimization

IEEE/CAA Journal of Automatica Sinica 2022年第5期9卷 812-833页

作者： Xinlei Yi Shengjun Zhang Tao Yang Tianyou Chai Karl Henrik Johansson Division of Decision and Control Systems School of Electrical Engineering and Computer ScienceKTH Royal Institute of Technologyand also affiliated with the Digital FuturesStockholm 10044Sweden Department of Electrical Engineering University of North TexasDentonTX 76203 USA IEEE State Key Laboratory of Synthetical Automation for Process Industries Northeastern UniversityShenyang 110819China

The distributed nonconvex optimization problem of minimizing a global cost function formed by a sum of n local cost functions by using local information exchange is *** problem is an important component of many machine learning techniques with data parallelism,such as deep learning and federated *** propose a distributed primal-dual stochastic gradient descent(SGD)algorithm,suitable for arbitrarily connected communication networks and any smooth(possibly nonconvex)cost *** show that the proposed algorithm achieves the linear speedup convergence rate O(1/(√nT))for general nonconvex cost functions and the linear speedup convergence rate O(1/(nT)) when the global cost function satisfies the Polyak-Lojasiewicz(P-L)condition,where T is the total number of *** also show that the output of the proposed algorithm with constant parameters linearly converges to a neighborhood of a global *** demonstrate through numerical experiments the efficiency of our algorithm in comparison with the baseline centralized SGD and recently proposed distributed SGD algorithms.

关键词： Distributed nonconvex optimization linear speedup Polyak-Lojasiewicz(P-L)condition primal-dual algorithm stochastic gradient descent

来源：评论

学校读者我要写书评

暂无评论

Automatic Skull Shape Completion of Defective Skulls Using Transformers for Cranial Implant Design

引用

Procedia computer science 2024年 235卷 3305-3314页

作者： Resmi S RimjhimPadam Singh Kannappan Palaniappan Department. of Computer Science and Engineering Amrita School of Computing Bengaluru Amrita VishwaVidyapeetham India Dept. of Electrical Engineering and Computer Science University of Missouri Missouri USA

Cranioplasty is a surgical method that restores the aesthetic and protecting function of a damaged skull by implanting material into the damaged *** and accurate design of patient specific cranial implants is very much required in the process of *** time consumption for designing and manufacturing of patient specific cranial implant has become an obstruction for cranioplasty procedures. Hence, a fully automatic and fast design of cranial implant becomes very important. The cranial implant design processmainly comprises of two steps. The former step concentrates on the automatic skull shape completion of defective skulls to fill the gaps and the cracks created in the skull. While the second step computes the difference between defective input and the completed skull for generating the implant. Currently computer aided design is used for the skull shape completion task which is a time consuming process. The application of deep learning techniques may result to faster and accurate skull shape completionwhich can be used for the design of patient specific cranial implants. This work proposes a novel approach combining 3D U-Net with Transformers for the automatic skull shape completion *** are using a vision transformer in the encoder section of the 3D U-Net architecture to consider the volumetric skull reconstruction as sequence- to-sequence prediction problem and to efficaciously grasp the global contextual information. The work also compares its performance with the famous variants of 3D U-Net deep network model namely, 3D U-Net and 3D U-Net with attention. From the values resulted for the dice coefficient metric it is clear that the proposed 3D U-Net with transformer approach performs better than the other two models on test images.

关键词： Cranioplasty 3DU-Net 3DU-Net with Attention 3DU-Net with Transformer Dice coefficient

来源：评论

学校读者我要写书评

暂无评论

Class-Aware Contextual Information for Semantic Segmentation 48

Class-Aware Contextual Information for Semantic Segmentation

引用

48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

作者： Tang, Huadong Zhao, Youpeng Jiang, Yingying Gan, Zhuoxin Wu, Qiang University of Technology Sydney School of Electrical and Data Engineering Australia China University of Central Florida Department of Computer Science United States

ISBN: (纸本)9781728163277

Exploring spatial contextual information is a well-adopted approach to achieving better semantic segmentation performance. However, most existing methods neglect the class association between the neighboring pixels. In this paper, we propose a CACINet, which consists of a Semantic Affinity Module (SAM) and a Class Association Module (CAM), to generate class-aware contextual information among pixels on a fine-grained level. SAM analyzes the affiliation of any two given pixels belonging to the same or different class. It produces intra-class and inter-class pixel contextual information. CAM classifies the image into different class regions globally and then it encodes the pixel based on the degree of affiliation of the pixels with each class in the image. In this way, it augments the class affiliation of the pixels into the corresponding context calculation. Comprehensive experiments demonstrate that the proposed method achieves competitive performance on two semantic segmentation benchmarks: ADE20K and PASCAL-Context. © 2023 IEEE.

关键词： class-aware contextual information semantic segmentation

来源：评论

学校读者我要写书评

暂无评论

ENTROPY IS NOT ENOUGH FOR TEST-TIME ADAPTATION: FROM THE PERSPECTIVE OF DISENTANGLED FACTORS 12

ENTROPY IS NOT ENOUGH FOR TEST-TIME ADAPTATION: FROM THE PER...

引用

12th International Conference on Learning Representations, ICLR 2024

作者： Lee, Jonghyun Jung, Dahuin Lee, Saehyung Park, Junsung Shin, Juhyeon Hwang, Uiwon Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of School of Computer Science and Engineering Soongsil University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of Division of Digital Healthcare Yonsei University Korea Republic of AIIS ASRI INMC ISRC Seoul National University Korea Republic of

Test-time adaptation (TTA) fine-tunes pre-trained deep neural networks for unseen test data. The primary challenge of TTA is limited access to the entire test dataset during online updates, causing error accumulation. To mitigate it, TTA methods have utilized the model output's entropy as a confidence metric that aims to determine which samples have a lower likelihood of causing error. Through experimental studies, however, we observed the unreliability of entropy as a confidence metric for TTA under biased scenarios and theoretically revealed that it stems from the neglect of the influence of latent disentangled factors of data on predictions. Building upon these findings, we introduce a novel TTA method named Destroy Your Object (DeYO), which leverages a newly proposed confidence metric named Pseudo-Label Probability Difference (PLPD). PLPD quantifies the influence of the shape of an object on prediction by measuring the difference between predictions before and after applying an object-destructive transformation. DeYO consists of sample selection and sample weighting, which employ entropy and PLPD concurrently. For robust adaptation, DeYO prioritizes samples that dominantly incorporate shape information when making predictions. Our extensive experiments demonstrate the consistent superiority of DeYO over baseline methods across various scenarios, including biased and wild. Project page is publicly available at https://***/DeYO/. © 2024 12th International Conference on Learning Representations, ICLR 2024. All rights reserved.

关键词： Entropy

来源：评论

学校读者我要写书评

暂无评论

Silicon-Based Charge Trapping Memory Devices for Next-Generation Flexible Electronics Application

IEEE Journal on Flexible Electronics

引用

IEEE Journal on Flexible Electronics 2023年第5期2卷 408-413页

作者： Das, Uttam Kumar Aslam, Nabeel Hussain, Muhammad Mustafa El-Atab, Nazek Program of ECE Computer Electrical Mathematical Science and Engineering Division Thuwal23955 Saudi Arabia KAUST Core Labs Thuwal23955 Saudi Arabia Purdue University Elmore Family School of ECE West LafayetteIN47907 United States

Silicon-based charge trapping memory (CTM) devices currently dominate the nonvolatile memory (NVM) technology due to their excellent reliability, scalability, and maturity of the manufacturing process;however, they are rigid, which limits their potential in futuristic applications, such as wearables and flexible displays, among many others;moreover, the majority of previously reported studies on flexible NVM have been limited to either nonsilicon-based CTM or to emerging NVM technologies, such as ferroelectric random access memory and memristors, among others, which still need to overcome different challenges, including, but not limited to, reliability issues and complementary metal-oxide-semiconductor (CMOS) compatibility. In this article, a flexible silicon-based CTM device is developed using a volumetric reduction process. A MOS memory structure with a 7-nm thin Si3N4 layer embedded between 5-nm tunneling (SiO2) and 9-nm blocking oxides (Al2O3) is first fabricated. Next, a backside sequential etching process transforms the rigid die into an ultrathin (20μm) flexible Si die for future application in flexible electronics. The fabricated two-terminal device exhibits a 2.32 V memory window in both rigid and bent states, as observed using capacitance-voltage measurements. A three-terminal version of the device was also developed, where the source and drain terminals were activated using an in situ doped plasma-enhanced CVD (PECVD) process. The flexible encapsulated chip was tested using current-voltage measurements, and the performance was compared at different mechanical bending conditions. The electrical characterization confirms that the flexible devices show similar endurance (104 cycles) and retention (105s) performances to their rigid counterparts. © 2022 IEEE.

关键词： Aluminum oxide

来源：评论

学校读者我要写书评

暂无评论

Early Fault Detection in Rolling Element Bearings Using Signal Processing Techniques 21

Early Fault Detection in Rolling Element Bearings Using Sign...

引用

21st IEEE International Conference on Smart Communities: Improving Quality of Life using AI, Robotics and IoT, HONET 2024

作者： Saeed, Ali Akram, M. Usman Khan, Muazzam A. Afsar, M. Belal Quaid-i-Azam University ICESCO Big Data Analytics and Edge Computing Department of Computer Science Islamabad Pakistan National University of Sciences and Technology Department of Computer and Software Engineering Islamabad Pakistan Otto-von-Guericke University Faculty of Electrical Engineering and Information Technology Maddeburg Germany

ISBN: (纸本)9798350378078

Many studies show that bearings are the most vulnerable components in low-voltage motors. While advanced bearing diagnostic systems exist, their cost can be a barrier for non-critical machinery due to the potential wait time to see a return on investment. This research explores the feasibility of using basic signal processing techniques on vibration data for bearing condition monitoring. The primary goal is to analyze benchmark data from the Case Western Reserve University (CWRU) dataset and establish a baseline performance for this data-driven approach. The results suggest that, with appropriate signal processing, it's possible to achieve early detection of bearing faults, leading to more efficient time-To-fault identification with higher detection accuracies ranging from 83 % to 100 % for each method. © 2024 IEEE.

关键词： Signal modulation

来源：评论

学校读者我要写书评

暂无评论

Active Space Debris Removal with Artificial Intelligence Assisted CubeSats using Robot Technology and Swarm Intelligence for Trajectory Prediction, Debris Capture, and Deorbiting in Low Earth Orbit 74

Active Space Debris Removal with Artificial Intelligence Ass...

引用

74th International Astronautical Congress, IAC 2023

作者： Vasudevan, Nijanthan Kavitha, Arjuna Karthikeyan Senthilvel Thach, Alex Massat, Oudayl Paoli, Cassandra Department of Electrical and Computer Engineering Drexel University PhiladelphiaPA United States Department of Computer Science University of Maryland MD United States Astrada Cyber Labs LLC PhiladelphiaPA United States

Low Earth Orbit (LEO) space debris buildup is a serious problem that puts important dangers to space missions and assets. The application of Artificial Swarm Intelligence (ASI), a new technology that can improve the efficiency of ADR operations, is a viable approach to solving this issue. This research suggests a unique method for ADR that employs robot equipped CubeSats with ASI support to collect and deorbit space debris. To enable coordinated and autonomous debris clearance, the proposed system combines several cutting-edge technologies, including swarm robotics, computer vision, and communication networks. The CubeSats can recognize, track, and collect space trash thanks to their sophisticated sensors and actuators, including robotic arms and cameras. To enable effective debris, capture and deorbiting, the ASI algorithm regulates the swarm's mobility, allowing it to respond to changes in the debris trajectory and surroundings. Realistic LEO settings and scenarios of the distribution and movement of debris are used in simulations to assess the performance of the proposed technology. The results demonstrate that in terms of effectiveness and thoroughness of debris removal, ASI-assisted CubeSats with robot technology beat conventional ADR techniques. The proposed technology has several potential advantages for the next space missions in addition to enhancing the security and sustainability of space operations. The system's flexibility and autonomy will allow for more effective and focused debris removal, improving the operational viability of ADR missions. Moreover, using CubeSats offers a scalable and affordable method for system deployment. The proposed system has restrictions that necessitate more study. The ASI algorithm's computational resource requirements are one restriction;however, this can be overcome with new hardware and methods. Future studies might potentially investigate the incorporation of other sensors and technology, such as robotics and machine

关键词： computer Vision CubeSats LEO Pattern Recognition Space debris Swarm Robotics

来源：评论

学校读者我要写书评

暂无评论

Spotted Hyena Optimizer with Deep Learning Driven Cybersecurity for Social Networks

引用

computer Systems science & engineering 2023年第5期45卷 2033-2047页

作者： Anwer Mustafa Hilal Aisha Hassan Abdalla Hashim Heba G.Mohamed Lubna A.Alharbi Mohamed K.Nour Abdullah Mohamed Ahmed S.Almasoud Abdelwahed Motwakel Department of Electrical and Computer Engineering International Islamic University MalaysiaKuala Lumpur53100Malaysia Department of Computer and Self Development Preparatory Year DeanshipPrince Sattam bin Abdulaziz UniversityAl-Kharj16278Saudi Arabia Department of Electrical Engineering College of EngineeringPrincess Nourah bint Abdulrahman UniversityP.O.Box 84428Riyadh11671Saudi Arabia Department of Computer Science College of Computers and Information TechnologyTabuk UniversityTabuk47512Saudi Arabia Department of Computer Sciences College of Computing and Information SystemUmm Al-Qura UniversityMecca24382Saudi Arabia Research Centre Future University in EgyptNew Cairo11845Egypt Department of Information Systems College of Computer and Information SciencesPrince Sultan UniversityRiyadh12435Saudi Arabia

Recent developments on Internet and social networking have led to the growth of aggressive language and hate *** provocation,abuses,and attacks are widely termed cyberbullying(CB).The massive quantity of user generated content makes it difficult to recognize *** advancements in machine learning(ML),deep learning(DL),and natural language processing(NLP)tools enable to detect and classify CB in social *** this view,this study introduces a spotted hyena optimizer with deep learning driven cybersecurity(SHODLCS)model for *** presented SHODLCS model intends to accomplish cybersecurity from the identification of CB in the *** achieving this,the SHODLCS model involves data pre-processing and TF-IDF based feature *** addition,the cascaded recurrent neural network(CRNN)model is applied for the identification and classification of ***,the SHO algorithm is exploited to optimally tune the hyperparameters involved in the CRNN model and thereby results in enhanced classifier *** experimental validation of the SHODLCS model on the benchmark dataset portrayed the better outcomes of the SHODLCS model over the recent approaches.

关键词： Cybersecurity cyberbullying online social network deep learning spotted hyena optimizer

来源：评论

学校读者我要写书评

暂无评论

Novel Aerial User Equipment Task Offloading Optimization in Integrated 6G Terrestrial and Non-Terrestrial Networks: A Deep Reinforcement Learning Approach

Novel Aerial User Equipment Task Offloading Optimization in ...

引用

2024 IEEE Global Communications Conference, GLOBECOM 2024

作者： Ei, Nway Nway Hassan, Sheikh Salman Tun, Yan Kyaw Han, Zhu Hong, Choong Seon Kyung Hee University Department of Computer Science and Engineering Yongin17104 Korea Republic of Aalborg University Department of Electronic Systems A. C. Meyers Vænge 15 København2450 Denmark University of Houston Department of Electrical and Computer Engineering HoustonTX77004 United States

ISBN: (纸本)9798350351255

This paper investigates a novel network architecture - the 6G integrated terrestrial-non-terrestrial network (ITNTN) with multi-access edge computing (ITNT-MEC). This system aims to bridge the connectivity gap between terrestrial infrastructure and non-terrestrial networks while offering real-time data processing through edge computing. We consider a scenario where aerial user equipments (AUEs) share resources of terrestrial base stations (TBSs) with terrestrial UEs (TUEs). We formulate an optimization problem to minimize the total energy consumption of both AUEs and TUEs. This problem involves joint optimization of AUE association (i.e., TBS or low Earth orbit (LEO) satellite), AUE trajectories, and TBS bandwidth allocation. Due to the dynamic network environment and non-convex optimization characteristics, solving this problem presents a significant challenge. To address this, we propose a novel algorithm that combines block coordinate descent (BCD) with deep deterministic policy gradient (DDPG) and a convex optimization method. Simulation results demonstrate the significant reductions in total energy consumption compared to baseline approaches, achieving improvements of 23%, 36.6%, and 46.5% against DQN-TO, RA, and FT, respectively. © 2024 IEEE.

关键词： Convex optimization

来源：评论

学校读者我要写书评

暂无评论

A Family of Digital T Workflows and Architectures: Exploring Two Cases 1

引用

1st International Conference on Innovative Intelligent Industrial Production and Logistics, IN4PL 2020 and 2nd International Conference on Innovative Intelligent Industrial Production and Logistics, IN4PL 2021

作者： Paredis, Randy Gomes, Cláudio Vangheluwe, Hans Department of Computer Science University of Antwerp Middelheimlaan 1 Antwerp Belgium DIGIT Department of Electrical and Computer Engineering Aarhus University Åbogade 34 Aarhus N Denmark Flanders Make@UAntwerp Antwerp Belgium

ISBN: (数字)9783031372285

ISBN: (纸本)9783031372278

Digital Models/Shadows/Twins/..have been given numerous definitions and descriptions in the literature. There is no consensus on terminology, nor a comprehensive description of workflows nor architectures. In this paper, we use the catch-all "Digital T" (pronounced "Digital Twinning") to refer to all concepts, techniques, architectures,..related to the "twinning" paradigm. In this paradigm, virtual instances, known as twins, of a System under Study (SuS) are continually updated with the SuS’s health, performance, and maintenance status, over its entire life-cycle. Digital T can be used for monitoring, analysis, optimization, and adaptation of complex engineered systems, in particular after these systems have been deployed. Digital T makes full use of both historical knowledge and of streaming data from sensors. Following Multi-Paradigm Modelling (MPM) principles, this paper proposes to explicitly model construction/use workflows as well as architectures and deployment of Digital T. Applying product family modelling allows for the de-/re-construction of the different Digital T variants in a principled, reproducible and partially automatable manner. Two small illustrative cases are discussed: a Line-Following Robot and an Incubator. These are representative for respectively an Automated Guided Vehicle and an Industrial Convection Oven, both important in an industrial context. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Life cycle

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：