检索结果-内蒙古大学图书馆

arXiv 2019年

作者： Sun, Qianru Liu, Yaoyao Chen, Zhaozheng Chua, Tat-Seng Schiele, Bernt School of Information Systems Singapore Management University School of Electrical and Information Engineering Tianjin University School of Computing National University of Singapore Department of Computer Vision and Machine Learning Max-Plank Institute for Informatics

Meta-learning has been proposed as a framework to address the challenging few-shot learning setting. The key idea is to leverage a large number of similar few-shot tasks in order to learn how to adapt a base-learner to a new task for which only a few labeled samples are available. As deep neural networks (DNNs) tend to overfit using a few samples only, typical meta-learning models use shallow neural networks, thus limiting its effectiveness. In order to achieve top performance, some recent works tried to use the DNNs pre-trained on large-scale datasets but mostly in straight-forward manners, e.g., (1) taking their weights as a warm start of meta-training, and (2) freezing their convolutional layers as the feature extractor of base-learners. In this paper, we propose a novel approach called meta-transfer learning (MTL) which learns to transfer the weights of a deep NN for few-shot learning tasks. Specifically, meta refers to training multiple tasks, and transfer is achieved by learning scaling and shifting functions of DNN weights for each task. In addition, we introduce the hard task (HT) meta-batch scheme as an effective learning curriculum that further boosts the learning efficiency of MTL. We conduct few-shot learning experiments and report top performance for five-class few-shot recognition tasks on three challenging benchmarks: miniImageNet, tieredImageNet and Fewshot-CIFAR100 (FC100). Extensive comparisons to related works validate that our MTL approach trained with the proposed HT meta-batch scheme achieves top performance. An ablation study also shows that both components contribute to fast convergence and high accuracy. Compared to the conference version of the paper [1], this version additionally presents (1) the analysis of using different DNN architectures, e.g., ResNet-12, ResNet-18 and ResNet-25;(2) the new MTL variants that adapt our scaling and shifting functions to the classical supervised and the state-of-the-art semi-supervised meta-learners, and

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Enterprise Architecture Trends in the Digital Transformation Era

Enterprise Architecture Trends in the Digital Transformation...

引用

International Seminar on Application for Technology of Information and Communication (ISemantic)

作者： Irfan Mahendra Ford Lumban Gaol Suhono Harso Supangkat Benny Ranti Computer Science Bina Nusantara University Jakarta Indonesia Faculty of Information Technology Nusa Mandiri University Jakarta Indonesia School of Electrical Engineering and Informatics Bandung Institute of Technology Bandung Indonesia Faculty of Computer Science University of Indonesia Jakarta Indonesia

Enterprise Architecture (EA) is a solution to build alignment between business strategy and information technology in dealing with digital transformation that causes fundamental changes for companies and businesses. The purpose of this study is to examine the trends and progress of EA in responding to the challenges of the digital transformation era, which is carried out through SLRs using the PRISMA method. According to the research findings, implementing EA in businesses serves the following purposes: establishing clear objectives and strategies for the development of business and technology, achieving harmony between business and information technology, time and cost efficiency, and effectiveness in business development processes and information technology. Then, it was concluded that there are nine components of EA, namely, vision, mission, and strategic objectives; business architecture; data & information architecture; information system architecture; infrastructure and network architecture; security architecture; human resources architecture; governance, legal, and compliance; and standards & regulations.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Cognitive Complexity of Software and Its Measurement

Cognitive Complexity of Software and Its Measurement

引用

Firth IEEE International Conference on Cognitive informatics(第五届认知信息国际会议)

作者： Yingxu Wang International Center for Cognitive Informatics (ICfCI) Dept of Electrical and Computer Engineering University of Calgary 2500 University Drive NW Calgary Alberta Canada T2N 1N4

ISBN: (纸本)1424404754;9781424404759

The estimation and measurement of functional complexity of software are an age-long problem in software engineering. The cognitive complexity of software presented in this paper is a new measurement for cross-platform analysis of complexities, sizes, and comprehension effort of software specifications and implementations in the phases of design, implementation, and maintenance in software engineering. This paper reveals that the cognitive complexity of software is a product of its architectural and operational complexities on the basis of deductive semantics and the abstract system theory. Ten fundamental basic control structures (BCS's) are elicited from software architectural and behavioral specifications and descriptions. The cognitive weights of those BCS's are derived and calibrated via a series of psychological experiments. Based on mis work, the cognitive complexity of software systems can be rigorously and accurately measured and analyzed Comparative case studies demonstrate that the cognitive complexity is highly distinguishable in software functional complexity and size measurement in software engineering.

关键词： Cognitive informatics software engineering BCS cognitive weights cognitive complexity psychological experiments calibrated weights cognitive functional size.

来源：评论

学校读者我要写书评

暂无评论

Z-upscaling: Optical Flow Guided Frame Interpolation for Isotropic Reconstruction of 3D EM Volumes

arXiv

引用

arXiv 2024年

作者： Ferede, Fisseha A. Khalighifar, Ali John, Jaison Venkataraman, Krishnan Khairy, Khaled Center for Bioimage Informatics St. Jude Children's Research Hospital MemphisTN38105 United States Department of Electrical and Computer Engineering University of Memphis MemphisTN38152 United States

We propose a novel optical flow based approach to enhance the axial resolution of anisotropic 3D EM volumes to achieve isotropic 3D reconstruction. Assuming spatial continuity of 3D biological structures in well aligned EM volumes, we reasoned that optical flow estimation techniques, often applied for temporal resolution enhancement in videos, can be utilized. Pixel-level motion is estimated between neighboring 2D slices along z, using spatial gradient flow estimates to interpolate and generate new 2D slices resulting in isotropic voxels. We leverage recent state-of-the-art learning methods for video frame interpolation and transfer learning techniques, and demonstrate the success of our approach on publicly available ultrastructure EM volumes. Copyright © 2024, The Authors. All rights reserved.

关键词： Optical flows

来源：评论

学校读者我要写书评

暂无评论

Perceptual Depth Quality Assessment of Stereoscopic Omnidirectional Images

arXiv

引用

arXiv 2024年

作者： Zhou, Wei Wang, Zhou The School of Computer Science and Informatics Cardiff University CardiffCF24 4AG United Kingdom The Department of Electrical and Computer Engineering University of Waterloo WaterlooONN2L 3G1 Canada

Depth perception plays an essential role in the viewer experience for immersive virtual reality (VR) visual environments. However, previous research investigations in the depth quality of 3D/stereoscopic images are rather limited, and in particular, are largely lacking for 3D viewing of 360-degree omnidirectional content. In this work, we make one of the first attempts to develop an objective quality assessment model named depth quality index (DQI) for efficient no-reference (NR) depth quality assessment of stereoscopic omnidirectional images. Motivated by the perceptual characteristics of the human visual system (HVS), the proposed DQI is built upon multi-color-channel, adaptive viewport selection, and interocular discrepancy features. Experimental results demonstrate that the proposed method outperforms state-of-the-art image quality assessment (IQA) and depth quality assessment (DQA) approaches in predicting the perceptual depth quality when tested using both single-viewport and omnidirectional stereoscopic image databases. Furthermore, we demonstrate that combining the proposed depth quality model with existing IQA methods significantly boosts the performance in predicting the overall quality of 3D omnidirectional images. © 2024, CC BY.

关键词： Stereo image processing

来源：评论

学校读者我要写书评

暂无评论

Prototype Generation with the sUpervised Classifier System on kNN Matching

Prototype Generation with the sUpervised Classifier System o...

引用

Congress on Evolutionary Computation

作者： Naoya Yatsu Hiroki Shiraishi Hiroyuki Sato Keiki Takadama Department of Informatics The University of Electro-Communications Tokyo Japan Department of Electrical Engineering and Computer Science Yokohama National University Yokohama Japan Interdisciplinary Information Science Research Division University of Tokyo Chiba Japan

ISBN: (数字)9798350308365

ISBN: (纸本)9798350308372

This paper focuses on “prototypes” as representative samples that can serve as a summary of the original dataset and proposes a novel Learning Classifier System (LCS) to generate a small number of prototypes that can achieve high classification accuracy in class classification. Concretely kNNUCS-PG is proposed to improve the sUpervised Classifier System (UCS) as one of LCSs by changing from the interval (rule) representation to the point representation to represent a prototype by employing k-nearest neighbors (kNN) to match the neighborhood prototypes. For a high classification accuracy with data reduction, selecting and generating prototypes that can maintain decision boundaries is necessary. Since UCS has generalization mechanisms, including rule merging, unnecessary rule deletion, and new rule discovery using niche Genetic Algorithm (GA), such techniques are highly useful for evolving indispensable prototypes as generalized prototypes. This paper compared conventional prototype-based methods with kNNUCS-PG in experiments using three benchmark problems and eight table datasets. kNNU CS-PG can maintain high classification accuracy even for complex problems with much overlap in the datasets. The results show that kNNUCS-PG is a robust PG that can maintain high classification accuracy while reducing data for various complex problems.

关键词： Accuracy Image color analysis Merging Supervised learning Prototypes Nearest neighbor methods Compaction

来源：评论

学校读者我要写书评

暂无评论

A Scoping Review on Quantization Methods for Medical Imaging AI

A Scoping Review on Quantization Methods for Medical Imaging...

引用

IEEE International Symposium on Biomedical Imaging

作者： Atika Rahman Paddo Ria Treesa Raju Judy Wawira Gichoya Saptarshi Purkayastha Dept. of Electrical & Computer Engineering Purdue University Indianapolis Indiana USA Dept. of BioHealth Informatics Indiana University Indianapolis Indiana USA Dept. of Radiology Emory University Georgia USA

ISBN: (数字)9798350313338

ISBN: (纸本)9798350313345

Deep neural networks with state-of-the-art (SOTA) performance on medical imaging datasets such as Medical Resonance Imaging (MRI), Ultrasound (US), and Computed Tomography (CT) requires resource intensive environments (computing speed, data size, power usage, and memory size). With continuing Moore's law of doubling computing resources on a chip, some might believe that large deep neural networks are appropriate. However, there are many low-resource settings where such large models are wasteful or not implementable at the point of care. In this paper, we review recent (2017-2023) papers to identify the impact of compressing high-performance, critical task imaging models. In the literature, we found that Convolution Neural Network (CNN)-based models performed similarly to full-precision models by re-training or quantization-aware training. 3D-CNN models show SOTA performance on volumetric data. However, few published papers quantized 3D-CNN models. We summarize that quantized models are useful for segmenting and classifying medical images and, thus, important for low-resource settings.

关键词： Analytical models Quantization (signal) Reviews Computed tomography Biological system modeling Artificial neural networks Predictive models

来源：评论

学校读者我要写书评

暂无评论

Development of CYMETEO Strategic Meteorological Infrastructure and Service in Cyprus

Development of CYMETEO Strategic Meteorological Infrastructu...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Christina Oikonomou Haris Haralambous Filippos Tymvios Demetris Charalambous Margarita Satraki Vassiliki Kotroni Konstantinos Lagouvardos Eleftherios Loizou Frederick Research Center Nicosia Cyprus Cloudwater Ltd Nicosia Cyprus Department of Electrical Engineering Computer Engineering and Informatics Nicosia Frederick University Cyprus Cyprus Department of Meteorology Nicosia Cyprus National Observatory of Athens Athens Greece Nicosia Development Agency Nicosia Cyprus

ISBN: (数字)9798350360325

ISBN: (纸本)9798350360332

This study aims to describe the main objectives and activities of the research project "Cyprus GNSS (Global Navigation Satellite System) Meteorology Enhancement" (CYGMEN) funded (€1.500.000) by the Cyprus Research Innovation Foundation (RIF) in the frames of "Strategic Research Infrastructures" Call for Proposals. CYGMEN was initiated at January 2024 with the aim to establish a Meteorological cluster (CyMETEO) in Cyprus that will strategically augment existing Frederick Research Center (FRC) and Department of Meteorology (DoM) infrastructure, through the introduction of: a) a Lighting detection network, b) a dense GNSS network for atmospheric water vapor estimation (supported by Cloudwater Ltd (CW) and Nicosia Development Agency (ANEL), c) a Radar Wind Profiler (RWP) and d) a Microwave Radiometer (MWR). Within this framework, the evaluation of the CyMETEO GNSS tropospheric analysis component prior to its deployment at Nicosia district is pursued in this study. Furthermore, CyMETEO will enrich severe weather forecasting, through the assimilation of relevant datasets into WRF Numerical Weather Prediction model (NWP), in the context of a near real-time local weather service, that will be realized through the key collaboration with the Meteo Unit of the National Observatory of Athens.

关键词： Global navigation satellite system Wind Technological innovation Spaceborne radar Weather forecasting Radar detection Predictive models

来源：评论

学校读者我要写书评

暂无评论

Automated nomenclature labeling of the bronchial tree in 3D-CT Lung images 5th

引用

5th International Conference on Medical Image Computing and computer-Assisted Intervention, MICCAI 2002

作者： Kitaoka, Hiroko Park, Yongsup Tschirren, Juerg Reinhardt, Joseph Sonka, Milan McLennan, Goeffrey Hoffman, Eric A. Division of Physiologic Imaging Dept of Radiology College of Medicine University of Iowa 200 Hawkins Drive Iowa CityIA52242 United States Dept of Informatics and Mathematical Science Graduate School of Engineering Science Osaka University 2-2 Yamadaoka SuitaOsaka363-0871 Japan Dept of Electrical and Computer Engineering College of Engineering University of Iowa 1402 SC Iowa CityIA52242 United States Dept of Biomedical Engineering College of Engineering University of Iowa 1402 SC Iowa CityIA52242 United States Dept of Internal Medicine College of Medicine University of Iowa 200 Hawkins Drive Iowa CityIA52242 United States Biomedical Physics Laboratory Brussels Free University Campus Erasme cp 613/3 808 Route de Lennik Brussels1070 Belgium

ISBN: (纸本)3540442251

A nomenclature labeling algorithm for the human bronchial tree down to sub-lobar segments is proposed, as a means of inter and intra subject comparisons for the evaluation of lung structure and function. The algorithm is a weighted maximum clique search of an association graph between a reference tree and an object tree. The adjacency between nodes in the association graph is defined so as to reflect the consistency between the bronchial name in the reference tree and the node connectivity in the object tree. Nodes in the association graph are weighted according to the similarity between two tree nodes in the respective trees. This algorithm is robust to various branching patterns and false branches that arise during segmentation processing. Experiments have been performed for nine airway trees extracted automatically from clinical 3D-CT data, where approximately 250 branches were contained. Of these, 95 % were accurately named. © Springer-Verlag Berlin Heidelberg 2002.

关键词： Terminology

来源：评论

学校读者我要写书评

暂无评论

Runtime Verification of Autonomous Systems Utilizing Digital Twins as a Service

Runtime Verification of Autonomous Systems Utilizing Digital...

引用

Autonomic Computing and Self-Organizing Systems Companion (ACSOS-C), IEEE International Conference on

作者： Morten Haahr Kristensen Alberto Bonizzi Cláudio Gomes Simon Thrane Hansen Carlos Isasa Hannes Iven Eduard Kamburjan Peter Gorm Larsen Martin Leucker Prasad Talasila Valdemar Trøjgård Tang Stefano Tonetta Lars B. Vosteen Thomas Wright Department of Electrical and Computer Engineering Aarhus University Denmark Fondazione Bruno Kessler Italy Interdisciplinary Centre for Security Reliability and Trust University of Luxembourg Institute for Software Engineering and Programming Languages Universität zu Lübeck Germany Department of Informatics University of Oslo Norway

ISBN: (数字)9798350389760

ISBN: (纸本)9798350389777

Autonomous Systems (AS) enable systems to adapt to drastic and unprecedented environmental changes, a capability that can be enhanced through the utilization of Digital Twins (DTs). However, the additional capabilities of AS come at the cost of explainability, as the expanding adaptation space complicates the reasoning about the system's behavior. For certain types of systems, it is crucial to ensure that specific properties are upheld despite the system's autonomous behavior. To facilitate the monitoring of these properties, we propose the use of Runtime Verification (RV). This tutorial demonstrates the integration of RV tools into the Digital Twins as a Service (DTaaS) platform to monitor and verify the behavior of AS in real-time. By exploring various methods to incorporate RV tools within a DT context, the tutorial aims to advance the application of RV technologies in autonomic computing and self-adaptive system design. Specifically, we demonstrate how the behavior of a self-configuring DT can be verified utilizing RV. This is accomplished through the DTaaS platform, which supports seamless deployment of DT-based AS.

关键词： Runtime Costs Autonomous systems Tutorials Real-time systems Cognition Digital twins Distributed computing Monitoring System analysis and design

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：