检索结果-内蒙古大学图书馆

Computers in Cardiology (CinC)

作者： Jefferson Dionisio Che Lin Lian-Yu Lin Wen-Chau Wu Smart Medicine and Health Informatics Program National Taiwan University (NTU) Taipei Taiwan Department of Electrical Engineering NTU Graduate Institute of Communication Engineering NTU School of Medicine NTU Center for Advanced Computing and Imaging in Biomedicine NTU Center for Biotechnology NTU Department of Internal Medicine NTU Hospital Taipei Taiwan Institute of Medical Device and Imaging NTU Department of Radiology College of Medicine NTU Graduate Institute of Clinical Medicine NTU

As part of the 2023 PhysioNet Challenge, our team FINDING_MEMO utilized Transformer to predict outcomes using patient EEG data since it excels at dealing with sequential data like EEG. We mainly used the Transformer encoder block's multi-head self-attention to generate representations from the input and leverage several hidden layers to form the final prediction. Using the latest EEG from every patient, our team achieved the challenge score of 0.42 with the hidden validation set (ranked 36th out of 73 invited teams) and obtained a result of 0.37 (ranked 29th out of 36 qualified teams). Our results show a consistent performance across varying EEG recording durations in both the validation and test set. Our team also had the second-best score when evaluated, with only 12 hours of available recordings in the test set. Such promising results showcase the models' generalizability and clinical potential in predicting outcomes for comatose patients, especially for limited available EEG recordings.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Learning interface conditions in domain decomposition solvers 22

Learning interface conditions in domain decomposition solver...

引用

Proceedings of the 36th International Conference on Neural Information Processing Systems

作者： Ali Taghibakhshi Nicolas Nytko Tareq Zaman Scott MacLachlan Luke Olson Matthew West Mechanical Science and Engineering University of Illinois Urbana-Champaign Urbana IL Computer Science University of Illinois Urbana-Champaign Urbana IL Scientific Computing Program Memorial University of Newfoundland and Labrador St. John's NL Canada Mathematics and Statistics Memorial University of Newfoundland and Labrador St. John's NL Canada

ISBN: (纸本)9781713871088

Domain decomposition methods are widely used and effective in the approximation of solutions to partial differential equations. Yet the optimal construction of these methods requires tedious analysis and is often available only in simplified, structured-grid settings, limiting their use for more complex problems. In this work, we generalize optimized Schwarz domain decomposition methods to unstructured-grid problems, using Graph Convolutional Neural Networks (GCNNs) and unsupervised learning to learn optimal modifications at subdomain interfaces. A key ingredient in our approach is an improved loss function, enabling effective training on relatively small problems, but robust performance on arbitrarily large problems, with computational cost linear in problem size. The performance of the learned linear solvers is compared with both classical and optimized domain decomposition algorithms, for both structured- and unstructured-grid problems.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Accelerating GMRES with deep learning in real-time

arXiv

引用

arXiv 2021年

作者： Luna, Kevin Klymko, Katherine Blaschke, Johannes P. Program in Applied Mathematics The University of Arizona United States Computational Research Division Lawrence Berkeley National Laboratory United States National Energy Research Scientific Computing Center Lawrence Berkeley National Laboratory United States

GMRES is a powerful numerical solver used to find solutions to extremely large systems of linear equations. These systems of equations appear in many applications in science and engineering. Here we demonstrate a real-time machine learning algorithm that can be used to accelerate the time-to-solution for GMRES. Our framework is novel in that is integrates the deep learning algorithm in an in situ fashion: the AI-accelerator gradually learns how to optimizes the time to solution without requiring user input (such as a pre-trained data set). We describe how our algorithm collects data and optimizes GMRES. We demonstrate our algorithm by implementing an accelerated (MLGMRES) solver in Python. We then use MLGMRES to accelerate a solver for the Poisson equation - a class of linear problems that appears in may applications. Informed by the properties of formal solutions to the Poisson equation, we test the performance of different neural networks. Our key takeaway is that networks which are capable of learning non-local relationships perform well, without needing to be scaled with the input problem size, making them good candidates for the extremely large problems encountered in high-performance computing. For the inputs studied, our method provides a roughly 2× acceleration. Copyright © 2021, The Authors. All rights reserved.

关键词： Poisson equation

来源：评论

学校读者我要写书评

暂无评论

Opportunities and Challenges from Using Animal Videos in Reinforcement Learning for Navigation

arXiv

引用

arXiv 2022年

作者： Giammarino, Vittorio Queeney, James Carstensen, Lucas C. Hasselmo, Michael E. Paschalidis, Ioannis Ch. Division of Systems Engineering Boston University BostonMA02215 United States Center for Systems Neuroscience Department of Psychological and Brain Sciences Graduate Program for Neuroscience Boston University BostonMA02215 United States College of Engineering Hariri Institute for Computing & Computational Science & Engineering Boston University BostonMA02215 United States

We investigate the use of animal videos (observations) to improve Reinforcement Learning (RL) efficiency and performance in navigation tasks with sparse rewards. Motivated by theoretical considerations, we make use of weighted policy optimization for off-policy RL and describe the main challenges when learning from animal videos. We propose solutions and test our ideas on a series of 2D navigation tasks. We show how our methods can leverage animal videos to improve performance over RL algorithms that do not leverage such observations. © 2022, CC BY.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Effects of Spacer and Single-Charge Trap on Voltage Transfer Characteristics of Gate-All-Around Silicon Nanowire CMOS Devices and Circuits

Effects of Spacer and Single-Charge Trap on Voltage Transfer...

引用

IEEE Conference on Nanotechnology

作者： Sekhar Reddy Kola Yiming Li Narasimhulu Thoti Parallel and Scientific Computing Laboratory EECS International Graduate Program from National Chiao Tung University Hsinchu 300 Taiwan Parallel and Scientific Computing Laboratory Institute of Communications Engineering National Chiao Tung University Hsinchu 300 Taiwan Parallel and Scientific Computing Laboratory EECS International Graduate Program National Chiao Tung University Hsinchu 300 Taiwan

ISBN: (数字)9781728182643

ISBN: (纸本)9781728182650

We report the effects of the spacer and the single-charge trap (SCT) on the voltage transfer characteristics of cylindrical-shape gate-all-around (GAA) silicon (Si) nanowire (NW) metal-oxide-semiconductor field effect transistor (MOSFETs). We explore the impact of low-x spacer, high-x spacer, and dual spacer (DS) on electrical characteristics of the GAA Si NW MOSFET with a gate length of 10 nm. Compared with the nominal device (i.e., the device without spacer), the device with DS possesses 68.8% reduction on the normalized off-current and 29.4% increase on the normalized on-current for n- and p-type devices. Similarly, 21.1% and 3.38% improvements on the normalized high and low noise margins can be achieved for the GAA Si NW complementary metal-oxide-semiconductor (CMOS) circuit. Notably, the voltage transfer characteristics induced by the acceptor- and donor-type SCT for the CMOS circuit with DS possesses 2.64% and 3.82% enhancements for the normalized high and low noise margins compared with the nominal one.

关键词： Silicon Gallium arsenide Logic gates MOSFET Semiconductor device modeling Hafnium compounds Numerical models

来源：评论

学校读者我要写书评

暂无评论

FAME: A Lightweight Spatio-Temporal Network for Model Attribution of Face-Swap Deepfakes

引用

Expert Systems with Applications 2025年

作者： Wasim Ahmad Yan-Tsung Peng Yuan-Hao Chang Institute of Information Science Academia Sinica Taipei 115 Taiwan Social Networks and Human-Centred Computing Taiwan International Graduate Program Taipei 115 Taiwan Department of Computer Science National Chengchi University Taipei 116 Taiwan

The widespread emergence of face-swap Deepfake videos poses growing risks to digital security, privacy, and media integrity, necessitating effective forensic tools for identifying the source of such manipulations. Although most prior research has focused primarily on binary Deepfake detection, the task of model attribution determining which generative model produced a given Deepfake remains underexplored. In this paper, we introduce FAME (Fake Attribution via Multilevel Embeddings), a lightweight and efficient spatio-temporal framework designed to capture subtle generative artifacts specific to different face-swap models. FAME integrates spatial and temporal attention mechanisms to improve attribution accuracy while remaining computationally efficient. We evaluate our model on three challenging and diverse datasets, which include Deepfake Detection and Manipulation (DFDM), FaceForensics++ (FF++), and FakeAVCeleb (FAVCeleb). The evaluation results show that FAME consistently performs better than existing methods in both accuracy and runtime, highlighting its potential for deployment in real-world forensic and information security applications. The code and pretrained models will be made publicly available at: https://***/wasim004/FAME/ .

关键词：

来源：评论

学校读者我要写书评

暂无评论

An Efficient NLMS-based VLSI Architecture for Robust FECG Extraction and FHR Processing

An Efficient NLMS-based VLSI Architecture for Robust FECG Ex...

引用

IEEE International Conference on Electronics, Circuits and Systems (ICECS)

作者： Patrícia U. L. da Costa Guilherme Paim Leandro Rocha Eduardo da Costa Sérgio Almeida Sergio Bampi Graduate Program on Electronic Engineering and Computing - Catholic University of Pelotas (UCPel) Brazil Graduate Program on Microelectronics (PGMicro) - Federal University of Rio Grande do Sul (UFRGS) Brazil

ISBN: (数字)9781728160443

ISBN: (纸本)9781728160450

This work presents an efficient NLMS-based VLSI architecture to extract the fetal electrocardiogram (FECG) and detect the fetal heart rate (FHR), using the adaptive filter strategy. The efficient NLMS-based architecture herein investigated can robustly cancel the high-noised mother-related ECG signals, enabling the FHR measurement. We used the Improved Fetal Pan and Tompkins Algorithm (IFPTA) to detect fetal R-peak and calculate the FHR. Our NLMS-based VLSI architecture effectively detects the R-peaks in the extracted FECG with 93.2% accuracy with the only 2.4 mW of total power dissipation.

关键词： Fetal heart rate Very large scale integration Computer architecture Electrocardiography Hardware Thorax Power dissipation

来源：评论

学校读者我要写书评

暂无评论

Combining m=2 Multipliers and Adder Compressors for Power Efficient Radix-4 Butterfly

Combining m=2 Multipliers and Adder Compressors for Power Ef...

引用

Midwest Symposium on Circuits and Systems (MWSCAS)

作者： Guilherme Ferreira Leandro M. G. Rocha Eduardo Costa Sergio Bampi Graduate Program in Microelectronics (PGMICRO) - Federal University of Rio Grande do Sul (UFRGS) Brazil Graduate Program in Electronic Engineering and Computing - Catholic University of Pelotas (UCPel) Brazil

ISBN: (数字)9781728180588

ISBN: (纸本)9781728180595

In this work, we propose a power-efficient hardware architecture for 16-bit m Radix-4 DIT (Decimation in Time) butterfly using radix-2 m multipliers, with m=2, and 4-2 adder compressors. The multiplier uses both Wallace and Dadda schemes in the addition tree, with Carry-Select, Kogge Stone, and '+' operator from the tool internally. We used a method for a realistic power extraction with the standard delay format. The results show that our best-proposed Radix-4 butterfly saves up to 25% of power dissipation when compared with the original Radix-4 butterfly using the synthesis tool operators for both adders and multipliers. The optimized structure combines the m=2 Wallace-based multiplier, employing Kogge-Stone adder into its addition tree, and 4-2 adder compressors.

关键词： Adders Compressors Tools Computer architecture Power dissipation Capacitance Libraries

来源：评论

学校读者我要写书评

暂无评论

Artificial Neural Network Model for Steel Strip Tandem Cold Mill Power Prediction 3rd

Artificial Neural Network Model for Steel Strip Tandem Cold ...

引用

3rd International Conference on Applied Informatics, ICAI 2020

作者： de Oliveira, Danilo G. da Silva, Eliton M. Miranda, Fabiano J. F. Filho, José F. S. Parpinelli, Rafael S. Graduate Program in Applied Computing Santa Catarina State University – UDESC Joinville Brazil Department of Computer Science Santa Catarina State University – UDESC Joinville Brazil ArcelorMittal – Global Research and Development Group São Francisco do Sul Brazil

ISBN: (纸本)9783030617011

In the cold rolling of flat steel strips, electric energy consumption is one of the highest expenses. Predicting the power requirements according to the line and product conditions can significantly impact the energy cost and, thus, on the business’s profitability. This paper proposes predicting the power requirements of a tandem cold rolling mill of steel strips on a coil-to-coil base applying Artificial Neural Networks (ANN) as a uni-variate regression problem. The tests yielded an MSE of 300.39 kW or 2.4% and are better than the acceptable 5% error margin for this project, indicating that the ANN presented satisfactory results. The application of six full-month worth of data in the trained ANN model showed the excellent correlation of the ANN predictions with the measured data, leading to the conclusion that the system is ready for the deployment for daily use for line engineers. Overall, the results obtained show that the steel industry can highly benefit from Industry 4.0 and Artificial Intelligence technologies. © 2020, Springer Nature Switzerland AG.

关键词： Automotive industry

来源：评论

学校读者我要写书评

暂无评论

Social network users switching platforms behaviour: A proposal for research explorations using a mixed method approach 4

Social network users switching platforms behaviour: A propos...

引用

4th International Conference on Information Technology, Information Systems and Electrical Engineering, ICITISEE 2019

作者： Hiele, Timothy McBush Widjaja, Andree E. Haryani, Calandra Alencia Ming Chuan University Applied Computing Program Taipei Taiwan Universitas Pelita Harapan Information System Tangerang Indonesia

ISBN: (纸本)9781728151182

This paper proposes a novel research proposal to explore users switching platforms behaviour within the context of social network sites. The aim is to investigate the key factors essential to social network users' decision to switch platform. In this regard, a mixed methods approach of an exploratory-qualitative study (interviews), followed by a confirmatory-quantitative study (survey) will be proposed, hence so called a mix-method research design. The proposed study also specifically aims to contribute to Information Systems literature by adding value to the users' switching decision in social network sites platform, and extending the use of switching cost theory. In particular, by anticipating the salient factors affecting the social network sites users to switch platforms, as well as interrelationships using a mixed method approach. © 2019 IEEE.

关键词： Surveys

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：