检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Jang, Hojun Kim, Young Min Dept. of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence and INMC Seoul National University Korea Republic of

We present Reusable Motion prior (ReMP), an effective motion prior that can accurately track the temporal evolution of motion in various downstream tasks. Inspired by the success of foundation models, we argue that a robust spatio-temporal motion prior can encapsulate underlying 3D dynamics applicable to various sensor modalities. We learn the rich motion prior from a sequence of complete parametric models of posed human body shape. Our prior can easily estimate poses in missing frames or noisy measurements despite significant occlusion by employing a temporal attention mechanism. More interestingly, our prior can guide the system with incomplete and challenging input measurements to quickly extract critical information to estimate the sequence of poses, significantly improving the training efficiency for mesh sequence recovery. ReMP consistently outperforms the baseline method on diverse and practical 3D motion data, including depth point clouds, LiDAR scans, and IMU sensor data. Project page is available in https://***/ReMP. Copyright © 2024, The Authors. All rights reserved.

关键词： Reusability

来源：评论

学校读者我要写书评

暂无评论

Exploring the Benefits of Leveraging AI for Enhancing Big Data Retrieval Performance 1

Exploring the Benefits of Leveraging AI for Enhancing Big Da...

引用

1st IEEE International Conference on Emerging Research in Computational science, ICERCS 2023

作者： Sharma, Neeraj Kaur, Ashmeet Murali, M.J. Sakhare, Sachin R. Singh, Vaishali Haroon, Syed Vivekananda Global University Department of Electrical Engineering Jaipur India Chitkara University Institute of Engineering and Technology Chitkara University Centre of Interdisciplinary Research in Business and Technology Punjab India Prince Shri Venkateshwara Padmavathy Engineering College Department of Electrical and Electronics Engineering Chennai India Vishwakarma Institute of Information Technology Department of Computer Engineering Pune India Maharishi University of Information Technology Maharishi School of Engineering and Technology Uttar Pradesh India Department of Computer Science and Information Technology Karnataka Bangalore India

ISBN: (纸本)9798350359756

AI algorithms were used to improve massive information retrieval overall performance as they lessen the quantity of manual work involved in data processing and retrieval. These algorithms use numerous device studying (ML) strategies along with supervised and unsupervised learning, as well as deep gaining knowledge of, to procedure, analyze, and interpret big datasets. Device learning (ML) models can be used to manner the records and choose out the maximum essential functions, that are then used to create predictive models to improve large facts retrieval performance. The predictive fashions can discover styles and anomalies within the facts that may be used to form better seek queries and enhance results. This allows lessen the time taken for statistics retrieval and might cause better accuracy and performance. additionally, AI algorithms may be used to lessen storage necessities due to the fact they can organization redundant statistics, thereby enhancing seek *** fashions can also be used to become aware of styles which might be predictive but which may not be obvious from a records set. This may similarly improve search overall performance, as the styles may be used to create tailored seek requests and tailor search results in order that they may be more relevant to the user. © 2023 IEEE.

关键词： Big data

来源：评论

学校读者我要写书评

暂无评论

Dynamic Mesh Recovery from Partial Point Cloud Sequence

Dynamic Mesh Recovery from Partial Point Cloud Sequence

引用

International Conference on computer Vision (ICCV)

作者： Hojun Jang Minkwan Kim Jinseok Bae Young Min Kim Dept. of Electrical and Computer Engineering Seoul National University Interdisciplinary Program in Artificial Intelligence and INMC Seoul National University

The exact 3D dynamics of the human body provides crucial evidence to analyze the consequences of the physical interaction between the body and the environment, which can eventually assist everyday activities in a wide range of applications. However, optimizing for 3D configurations from image observation requires a significant amount of computation, whereas real-world 3D measurements often suffer from noisy observation or complex occlusion. We resolve the challenge by learning a latent distribution representing strong temporal priors. We use a conditional variational autoencoder (CVAE) architecture with a transformer to train the motion priors with a large-scale motion dataset. Then our feature follower effectively aligns the feature spaces of noisy, partial observation with the necessary input for pre-trained motion priors, and quickly recovers a complete mesh sequence of motion. We demonstrate that the transformer-based autoencoder can collect necessary spatio-temporal correlations robust to various adversaries, such as missing temporal frames, or noisy observation under severe occlusion. Our framework is general and can be applied to recover the full 3D dynamics of other subjects with parametric representations.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Regional Wind Power Forecasting Based on CNNBiLSTM-ReliefF Method

Regional Wind Power Forecasting Based on CNNBiLSTM-ReliefF M...

引用

Iranian Conference on Renewable Energy and Distributed Generation (ICREDG)

作者： Fariba Dehghan Mohsen Parsa Moghaddam Maryam Imani Faculty of Interdisciplinary Science and Technology Tarbiat Modares University Tehran Iran Faculty of Electrical and Computer Engineering Tarbiat Modares University Tehran Iran

ISBN: (数字)9798350361322

ISBN: (纸本)9798350361339

The noisy and stochastic nature of wind power production introduces many challenges to the grid. Wind power forecasting methods can effectively predict these uncertainties and ensure the stable operation of power systems. Due to the superior performance of deep learning approaches in predicting nonlinear data, a hybrid method based on two-dimensional convolutional neural network (Conv2D), bidirectional long short-term memory (BiLSTM), and ReliefF algorithm is presented in this study. First, wind power data are preprocessed. Then, the processed data are entered into each Conv2D and BiLSTM network for temporal-spatial feature extraction. Finally, the features that are extracted from these two networks are combined by assigning weights to them through the ReliefF algorithm for final wind power prediction. The forecasting performance of the presented method is tested on a real aggregated dataset of Belgium for 15 minutes ahead regional wind power forecasting. According to the results, the proposed model has higher accuracy than long short-term memory, gated recurrent unit (GRU), BiLSTM, and Conv2D networks.

关键词： Training Recurrent neural networks Accuracy Uncertainty Wind power generation Feature extraction Prediction algorithms

来源：评论

学校读者我要写书评

暂无评论

Sleep Monitoring Systems based on Edge Computing and Microservices Caching

Sleep Monitoring Systems based on Edge Computing and Microse...

引用

Artificial Intelligence of Things (AIoT), IEEE Annual Congress on

作者： Nico Surantha David Jayaatmaja Sani Muhamad Isa Faculty of Science and Engineering Department of Electrical Electronics and Communication Engineering Tokyo City University Tokyo Japan Computer Science Department BINUS Graduate Program-Master of Computer Science Bina Nusantara University Jakarta Indonesia

ISBN: (数字)9798350392296

ISBN: (纸本)9798350392302

The rapid development of Internet of Things (IoT) technology has enabled the widespread deployment of health monitoring systems. Traditionally, the health monitoring system has been limited by centralized processing and storage in the cloud, leading to latency issues and potential data loss. This paper introduces a smart sleep monitoring system based on edge computing, utilizing microservices architecture and caching techniques. The proposed system employs edge computing to enable data processing closer to the source, reducing latency and improving real-time monitoring capabilities. Caching is employed to reduce database load and optimize random access memory (RAM) usage. This research addresses latency and response time challenges on IoT health monitoring platforms in environments with poor network quality while optimizing database load and resource usage on Jetson Nano as the edge computing device. Using Electrocardiogram (ECG) data as input, the proposed system yields impressive performance metrics. The research results indicate that the proposed system can increase throughput by 26.92 KB/s, reduce response time by 18.8 ms, and decrease latency by 20.86 ms compared to the previous work. Message Queuing Telemetry Transport (MQTT) integration reduces CPU usage by approximately 40% and RAM usage by about 81.24%.

关键词： Cloud computing Databases Random access memory Microservice architectures Real-time systems Nanoscale devices Internet of Things

来源：评论

学校读者我要写书评

暂无评论

A Cost-Sensitive Genetic Algorithm for Customer Segmentation in Auto Insurances

A Cost-Sensitive Genetic Algorithm for Customer Segmentation...

引用

International eConference on computer and Knowledge engineering (ICCKE)

作者： Alireza Khajenoori Mohammad Saniee Abadeh Mohsen Mohammadzadeh Faculty of Interdisciplinary Science and Technology Tarbiat Modares University Tehran Iran Faculty of Electrical and Computer Engineering Tarbiat Modares University Tehran Iran

ISBN: (数字)9798331511272

ISBN: (纸本)9798331511289

In the auto insurance industry, accurate risk assessment is essential for determining fair premiums and managing claims effectively. As customer data grows in volume and complexity, machine learning has become crucial for precise customer segmentation and risk prediction. However, the industry faces significant challenges, particularly the class imbalance in data and the unequal costs associated with misclassification errors, where some errors are far more costly than others. This study compares two approaches to address these issues: traditional sampling techniques commonly used to mitigate class imbalance, and a cost-sensitive learning framework that optimizes the cost matrix using a genetic algorithm to minimize the financial impact of misclassification. The findings demonstrate that the cost-sensitive approach significantly enhances overall model performance by more effectively prioritizing costly errors, leading to more accurate and economically sound decision-making. This research highlights the importance of integrating advanced machine learning techniques in the insurance sector, showing that such approaches can substantially improve the fairness and efficiency of risk prediction models, ultimately benefiting insurers by enabling more precise premium setting and increasing customer satisfaction.

关键词： Industries Costs Accuracy Scalability Decision making Insurance Machine learning Predictive models Risk management Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

Low-Resolution Face Recognition for CCTV and Edge-Powered Smart Attendance Systems

Low-Resolution Face Recognition for CCTV and Edge-Powered Sm...

引用

IEEE Annual International computer Software and Applications Conference (COMPSAC)

作者： Nico Surantha Edward Yose Sani Muhamad Isa Department of Electrical Electronics and Communication Engineering Faculty of Science and Engineering Tokyo City University Tokyo Japan Computer Science Department BINUS Graduate Program-Master of Computer Science Bina Nusantara University Jakarta Indonesia

ISBN: (数字)9798350376968

ISBN: (纸本)9798350376975

Attendance systems have become more modern, and one of the biometric systems without physical contact is face recognition. However, many face-based attendance systems still carry out attendance individually and cannot detect multiple faces simultaneously. In addition, capturing facial data in real-time is still a challenge because the relatively large distance between the camera and the individual reduces the ability to recognize faces. The general solution is to use super-resolution to generate better-quality faces while maintaining the main facial recognition features. One technique still being researched is super-resolution generative adversarial networks (SRGAN). SRGAN can enlarge the resolution of captured images and maintain image quality sufficient for face recognition. The attendance system can be easily integrated into edge devices such as the Jetson Nano. This paper proposes automatic and effective attendance systems with the super-resolution technique to detect and recognize faces in low-resolution input. The experimental results show that using face data capture with a resolution of 40 × 40 pixels and a four-fold magnification results in a resolution of 160 × 160 pixels. Combining Face SRGAN with FaceNet architecture as the basis of face recognition can achieve an accuracy rate of 78.19% and an F1-Score of 81.13% with an average processing time of 1.61 seconds per frame on a PC and 14.55 seconds per frame on a Jetson Nano at an average of face recognition per frame of as many as up to 8 faces simultaneously.

关键词： Support vector machines Image quality Accuracy Face recognition Image edge detection Superresolution Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Light extraction improvement via ITO p-electrodes for InGaN red micro-LEDs emitting at 640 nm

Optics Continuum

引用

Optics Continuum 2025年第5期4卷 1040-1050页

作者： Cesur Altinkaya Mohammed A Najmi Daisuke Iida Kazuhiro Ohkawa Material Science and Engineering Program Physical Sciences and Engineering Division King Abdullah University of Science and Technology (KAUST) Thuwal 23955-6900 Saudi Arabia cesur.altinkaya@kaust.edu.sa Electrical and Computer Engineering Program Computer Electrical and Mathematical Sciences and Engineering Division King Abdullah University of Science and Technology (KAUST) Thuwal 23955-6900 Saudi Arabia daisuke.iida@kaust.edu.sa kazuhiro.ohkawa@kaust.edu.sa

We present light extraction efficiency (LEE) improvement for InGaN red micro-light emitting diodes (micro-LEDs) of various sizes operating at low current densities. We compared the characteristics of micro-LEDs with indium tin oxide (ITO) transparent p-electrodes with conventional opaque metal p-electrodes. 50 µm × 50 µm micro-LEDs with ITO p-electrodes achieved a peak on-wafer external quantum efficiency (EQE) of 2.54% with an emission wavelength of 640 nm at a current density as low as 0.4 A/cm 2 . This represents a 1.18-fold improvement in peak EQE compared to devices with metal p-electrodes. Light ray tracing simulation confirmed that the ITO p-electrodes exhibit 1.18 times higher light escape than metal-based micro-LEDs, validating the role of enhanced light extraction. These findings provide valuable insights for advancing high-definition display and VR applications.

关键词： Atomic layer deposition Indium tin oxide Light extraction Opacity Quantum efficiency Ray tracing

来源：评论

学校读者我要写书评

暂无评论

Gain Enhancement of Antenna-on-Chip at 94 GHz with an Integrated Artificial Magnetic Conductor for 6G System-on-Chip

Gain Enhancement of Antenna-on-Chip at 94 GHz with an Integr...

引用

IFIP International Conference on Very Large Scale Integration (VLSI-SoC)

作者： Yiyang Yu Atif Shamim Electrical and Computer Engineering Program King Abdullah University of Science and Technology Thuwal Saudi Arabia

Silicon-based complementary metal oxide semiconductor (CMOS) process has become one of the most popular processes to realize system-on-chip (SoC). However, as one of the essential components of wireless SoC, antennas are typically suffering from the poor radiation because of the highly conductive silicon substrate. Such antennas are known as antenna-on-chip (AoC). To enhance the radiation performance of AoC, artificial magnetic conductors (AMC) with double periodic strip structure layers has been proposed in this paper that can not only provide in-phase reflection but also isolate the antenna from the lossy silicon substrate. The proposed AMC shows a gain enhancement of 4.5 dB. The AMC-backed AoC is well-matched within 77-125 GHz and provides a boresight gain of 2 dBi at 94 GHz.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Analysis of Gait Data Based on Deep Learning to Predict Multiple Balance Scale Scores

引用

IEEE Sensors Journal 2024年第1期24卷 920-930页

作者： Lin, Chueh-Ho Peng, Chih-Wei Lee, I-Jung Wu, Chi-Ming Lin, Bor-Shing Taipei Medical University International Ph.D. Program in Gerontology and Long-Term Care Taipei City110 Taiwan Taipei Medical University Research Center in Nursing Clinical Practice Wan Fang Hospital Taipei City110 Taiwan Taipei Medical University School of Biomedical Engineering College of Biomedical Engineering Taipei City110 Taiwan Taipei Medical University School of Gerontology and Long-Term Care College of Nursing Taipei City110 Taiwan National Taipei University Department of Computer Science and Information Engineering New Taipei City237 Taiwan National Taipei University College of Electrical Engineering and Computer Science New Taipei City237 Taiwan

Unbalanced walking is increasingly common among older adults;therefore, routinely assessing the balance of older adults is crucial. The traditional method of assessing balance uses scales, requires the supervision of a physical therapist (PT), and is time-consuming. The present study, therefore, proposes a deep learning (DL) model architecture that combines a convolutional neural network (CNN) and a long short-term memory (LSTM) network model to predict the scores of three scales, namely the Berg balance scale (BBS), timed up and go (TUG), and single-leg stance (SLS). The gait data of 15 m of walking were collected from seven inertial measurement units (IMUs) and input into the CNN-LSTM model for evaluation. The BBS and TUG only require the participants to wear two IMUs on the left and right thighs, respectively, for accurate predictions. The mean absolute errors (MAEs) of predicting the scores of the BBS and TUG are 1.2562 and 1.4016 s, respectively;however, the MAE of score predictions of SLS is higher than that of the BBS and TUG, indicating that gait data cannot be used for assessing SLS;moreover, participants only wear one IMU on the right calf for BBS and TUG evaluations, which yield MAEs of 1.4334 and 1.5229 s, respectively. The proposed system can quickly and accurately predict the scores of the BBS and TUG. The proposed model can assist PTs with making clinical decisions. © 2001-2012 IEEE.

关键词： Long short-term memory

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：