检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Li, Jiaxiang Chen, Xuxing Ma, Shiqian Hong, Mingyi Department of Electrical and Computer Engineering University of Minnesota Twin Cities United States Department of Mathematics University of California Davis United States Department of Computational Applied Math and Operations Research Rice University United States

Existing decentralized algorithms usually require knowledge of problem parameters for updating local iterates. For example, the hyperparameters (such as learning rate) usually require the knowledge of Lipschitz constant of the global gradient or topological information of the communication networks, which are usually not accessible in practice. In this paper, we propose D-NASA, the first algorithm for decentralized nonconvex stochastic optimization that requires no prior knowledge of any problem parameters. We show that D-NASA has the optimal rate of convergence for nonconvex objectives under very mild conditions and enjoys the linear-speedup effect, i.e. the computation becomes faster as the number of nodes in the system increases. Extensive numerical experiments are conducted to support our findings. Copyright © 2024, The Authors. All rights reserved.

关键词： NASA

来源：评论

学校读者我要写书评

暂无评论

LSTM-XGB: A New Deep Learning Model for Human Activity Recognition based on LSTM and XGBoost 7

LSTM-XGB: A New Deep Learning Model for Human Activity Recog...

引用

7th International Conference on Digital Arts, Media and Technology, DAMT 2022 and 5th ECTI Northern Section Conference on Electrical, Electronics, computer and Telecommunications engineering, NCON 2022

作者： Mekruksavanich, Sakorn Jantawong, Ponnipa Jitpattanakul, Anuchit University of Phayao School of Information and Communication Technology Department of Computer Engineering Phayao Thailand King Mongkut's University of Technology North Bangkok Faculty of Applied Science Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Bangkok Thailand

ISBN: (纸本)9781665495103

Recognition of human activity focuses on identifying various human motions and behaviors using data acquired from multiple types of sensors, a branch of computer science. Considering time series data, deep learning algorithms show promise compared to laborious handcrafted feature extraction methods, which are highly dependent on the quality of the used domain parameters. This paper proposes a novel deep learning model for HAR issues called the Long Short Term Memory Neural network with eXtreme Gradient Boosting (LSTM-XGB), developed on recurrent neural networks with eXtreme Gradient Boosting (XG-Boost). The LSTM-XGB comprises LSTM layers to understand the characteristics of the input and is capable of learning features automatically, followed by XGBoost in the final layer to forecast the class labels, which serves to enhance detection capability. When sufficient circumstances are satisfied, the LSTM-XGB model can be simpler by lowering the number of parameters. It is not essential to readjust the weighting factors throughout a backpropagation phase. Investigations using smartphone sensor data were carried out using a publicly available standardized HAR dataset called the UniMiB-SHAR dataset, which contains activities associated with everyday life and activities related to falls. The experiments show that the LSTM-XGB provides much superior identification capabilities than other standard deep learning techniques, with the highest accuracy of 92.59%. Furthermore, the LSTM-XGB models surpass the existing best-practice models on the same dataset. © 2022 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

Pre-Impact Fall Detection Based on Wearable Inertial Sensors using Hybrid Deep Residual Neural Network 6

Pre-Impact Fall Detection Based on Wearable Inertial Sensors...

引用

6th International Conference on Information Technology, InCIT 2022

作者： Mekruksavanich, Sakorn Jantawong, Ponnipa Jitpattanakul, Anuchit University of Phayao School of Information and Communication Technology Department of Computer Engineering Phayao Thailand King Mongkut's University of Technology North Bangkok Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Faculty of Applied Science Bangkok Thailand

ISBN: (纸本)9781665489126

Falls are the primary cause of fatal and nonfatal injuries among the elderly. Consequently, pre-impact fall detection that identifies a fall before the body's collision with the ground is of essential importance. In the latest days, academics have moved their attention from post-impact fall detection to pre-impact fall detection. Pre-impact fall detection systems typically use either a threshold-based or machine learning-based approach, and the thresholding is challenging to determine accurately for threshold-based methods. In addition, while additional features can sometimes assist in categorizing falls and non-falls more precisely, the value computation of prominent features will be too time-intensive, using too much of the algorithm's operating time. We proposed a pre-impact fall detection method employing wearable inertial sensors and a deep residual model to address the limitations of feature extraction, threshold definition, and algorithm sophistication. After training on a large-scale motion dataset known as the KFall, the suggested deep learning model could identify with 91.87% accuracy in 0.5 second. © 2022 IEEE.

关键词： Fall detection

来源：评论

学校读者我要写书评

暂无评论

High-dimensional inference for the average treatment effect under model misspecification using penalized bias-reduced double-robust estimation

引用

Biostatistics and Epidemiology 2022年第2期6卷 221-238页

作者： Avagyan, Vahe Vansteelandt, Stijn Biometris Wageningen University and Research Wageningen Wageningen Netherlands Department of Applied Mathematics Computer Science and Statistics Ghent University Gent Belgium Department of Medical Statistics London School of Hygiene and Tropical Medicine London United Kingdom

The presence of confounding by high-dimensional variables complicates the estimation of the average effect of a point treatment. On the one hand, it necessitates the use of variable selection strategies or more general high-dimensional statistical methods. On the other hand, the use of such techniques tends to result in biased estimators with a non-standard asymptotic behavior. Double-robust estimators are useful for offering a resolution because they possess a so-called small bias property. This property has been exploited to achieve valid (uniform) inference of the average causal effect when data-adaptive estimators of the propensity score and conditional outcome mean both converge to their respective truths at sufficiently fast rate. In this article, we extend this work in order to retain valid (uniform) inference when one of these estimators does not converge to the truth, regardless of which. This is done by generalizing prior work for low-dimensional settings by [Vermeulen K, Vansteelandt S. Bias-reduced doubly robust estimation. Am Stat Assoc. 2015;110(511):1024–1036.] to incorporate regularization. The proposed penalized bias-reduced double-robust estimation strategy exhibits promising performance in simulation studies and a data analysis, relative to competing proposals. © 2021 International Biometric Society–Chinese Region.

关键词： Confounding debiasing double robustness high-dimensional inference model misspecification penalized estimating equations

来源：评论

学校读者我要写书评

暂无评论

Innovative methods of signal processing for hydrogen-filled proportional counters

引用

IEEE Sensors Journal 2025年第11期25卷 19159-19167页

作者： Horna, Aleš Kolář, Ondřej Kubíček, Michal Czakoj, Tomáš Košťál, Michal Mazánková, Věra Petr, Jiří Cvachovec, František Ančář, Aleš Král, Jan Přenosil, Václav Matěj, Zdeněk Masaryk University Faculty of Informatics Department of Computer Systems and Communications Brno602 00 Czech Republic Brno University of Technology Faculty of Electrical Engineering and Communication Department of Radioelectronics Brno616 00 Czech Republic Research Centre Řež Husinec-Řež250 68 Czech Republic University of Defence Faculty of Military Technology Department of Mathematics and Physics Brno602 00 Czech Republic VF a.s. Černá Hora 679 21 Czech Republic

Proportional counters filled with hydrogen are suitable for fast neutron detection and spectrometry. As their sensitivity is not limited only to this type of ionizing radiation as desired, an approach suitable for event discrimination was investigated. Several counters of the given construction were subjected to various radiation fields and the effort was directed to the examination of pulse-shape discrimination method intended to distinguish neutrons from gamma rays and parasitic events (microdischarges) if these also occurred. Rising edge duration of the output pulses and their amplitudes were mapped into diagrams enabling visual discrimination while the criteria for further computational processes are to be determined. Pulse waveforms from charge-sensitive preamplifier with the proportional counters connected were recorded using an FPGA-based custom-build acquisition unit and then post-processed. Moreover, the possibility of operation verification and energy calibration based on the introduction of helium-3 admixture into the gas filling was studied, as well as the energy resolution abilities. Also, the theoretical basis for particle discrimination is presented in the form of description of events and principles involved. © 2001-2012 IEEE.

关键词： Gamma rays

来源：评论

学校读者我要写书评

暂无评论

Two-Way Communication Retrial Queue with Markov Modulated Poisson Input and Multiple Types of Outgoing Calls 25th

Two-Way Communication Retrial Queue with Markov Modulated P...

引用

25th International Conference on Distributed computer and Communication Networks, DCCN 2022

作者： Nazarov, Anatoly Phung-Duc, Tuan Paul, Svetlana Lizyura, Olga Institute of Applied Mathematics and Computer Science National Research Tomsk State University 36 Lenina ave. Tomsk634050 Russia Faculty of Engineering Information and Systems University of Tsukuba 1-1-1 Tennodai Ibaraki Tsukuba305-8573 Japan

ISBN: (纸本)9783031306471

In this paper, we consider single server two-way communication retrial queue with Markov modulated Poisson input. Such system serves two types of calls: incoming and outgoing. We assume that incoming calls arrive according to the input process and retry in case the server is busy upon arrival. If the server is occupied upon arrival, the incoming call joins the orbit and waits for some time before making the next try to receive service. On the other hand, outgoing calls are initiated by server and do not make redial. We present the analysis of the queue using asymptotic diffusion method. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Queueing theory

来源：评论

学校读者我要写书评

暂无评论

An Almost Singularly Optimal Asynchronous Distributed MST Algorithm 36

An Almost Singularly Optimal Asynchronous Distributed MST Al...

引用

36th International Symposium on Distributed Computing, DISC 2022

作者： Dufoulon, Fabien Kutten, Shay Moses, William K. Pandurangan, Gopal Peleg, David Department of Computer Science University of Houston HoustonTX United States Faculty of Industrial Engineering and Management Technion - Israel Institute of Technology Haifa Israel Department of Computer Science and Applied Mathematics Weizmann Institute of Science Rehovot Israel

ISBN: (纸本)9783959772556

A singularly (near) optimal distributed algorithm is one that is (near) optimal in two criteria, namely, its time and message complexities. For synchronous CONGEST networks, such algorithms are known for fundamental distributed computing problems such as leader election [Kutten et al., JACM 2015] and Minimum Spanning Tree (MST) construction [Pandurangan et al., STOC 2017, Elkin, PODC 2017]. However, it is open whether a singularly (near) optimal bound can be obtained for the MST construction problem in general asynchronous CONGEST networks. In this paper, we present a randomized distributed MST algorithm that, with high probability, computes an MST in asynchronous CONGEST networks and takes Õ(D1+Ε + √n) time and Õ(m) messages1, where n is the number of nodes, m the number of edges, D is the diameter of the network, and Ε > 0 is an arbitrarily small constant (both time and message bounds hold with high probability). Since (Equation presented)(D + √n) and Ω(m) are respective time and message lower bounds for distributed MST construction in the standard KT0 model, our algorithm is message optimal (up to a polylog(n) factor) and almost time optimal (except for a DΕ factor). Our result answers an open question raised in Mashregi and King [DISC 2019] by giving the first known asynchronous MST algorithm that has sublinear time (for all D = O(n1-Ε)) and uses Õ(m) messages. Using a result of Mashregi and King [DISC 2019], this also yields the first asynchronous MST algorithm that is sublinear in both time and messages in the KT1 CONGEST model. A key tool in our algorithm is the construction of a low diameter rooted spanning tree in asynchronous CONGEST that has depth Õ(D1+Ε) (for an arbitrarily small constant Ε > 0) in Õ(D1+Ε) time and Õ(m) messages. To the best of our knowledge, this is the first such construction that is almost singularly optimal in the asynchronous setting. This tree construction may be of independent interest as it can also be used for efficiently perfor

关键词： Distributed computer systems

来源：评论

学校读者我要写书评

暂无评论

Accuracy Improvement of Complex Sensor-based Activity Recognition Using Hybrid CNN 6

Accuracy Improvement of Complex Sensor-based Activity Recogn...

引用

6th International Conference on Information Technology, InCIT 2022

作者： Mekruksavanich, Sakorn Jantawong, Ponnipa Jitpattanakul, Anuchit University of Phayao Department of Computer Engineering School of Information and Communication Technology Phayao Thailand King Mongkut's University of Technology North Bangkok Faculty of Applied Science Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Bangkok Thailand

ISBN: (纸本)9781665489126

Applying smartphones and smartwatches for Human ActivitiesRecognition (HAR) is beneficial for analyzing the participant's dailybehavior and identifying mishaps, such as fall detection, enabling early healthcare monitoring and fall detection. Numerous HAR studies have investigated different recognition models based on sensor data. Nonetheless, this research mainly focused on basic motions such as sitting, standing and walking. In actual settings, human actions are often accomplished in complex ways. This research will examine the recognition of complicated activities using CNN and LSTM networks as deep learning classifiers. Furthermore, this study presented hybrid CNN for enhancing the identification of complicated activities. We evaluated these deep learning classifiers using WISDM-HARB, a publically available benchmark HAR. This dataset used accelerometer and gyroscope data from smartphones and smartwatches to collect sensor data of complicated activities. The experiment demonstrates that all deep learning classifiers recognize basic actions more accurately than complicated ones. Also, it was observed that the hybrid CNN classifier presented had the most fantastic accuracy with 96.25%. According to the research, these classifiers are effective at recognizing basic activities, but their performance decreases as the sophistication of the actions increases. Nonetheless, hybrid CNN models are competent for handling complicated tasks. © 2022 IEEE.

关键词： Smartphones

来源：评论

学校读者我要写书评

暂无评论

A Machine Learning and Ratemaking Evaluation of Four Auto Insurance Pure Premium Modeling Algorithms 28

A Machine Learning and Ratemaking Evaluation of Four Auto In...

引用

28th International computer Science and engineering Conference, ICSEC 2024

作者： Kumar, Rakesh Rakhra, Manik Prashar, Deepak Upadhyay, Saurabh Mrsic, Leo Khan, Arafat Ahmad Lovely Professional University Department of Mathematics and Statistics Punjab Phagwara India Lovely Professional University School of Computer Science and Engineering Punjab Phagwara India Algebra University Vice Rector for Science and Research Zagreb10000 Croatia Khon Kaen Thailand

ISBN: (纸本)9798350366860

Actuarial science is increasingly using machine learning. Machine learning has improved actuarial pricing prediction by forecasting future claims. Insurers are testing these algorithms, but model explainability and implementation cost are issues. The current literature barely touches upon this broad field of inquiry. Fujita et al. published a 2020 study on predictive frequency models using AGLM, GLM, GAM, and GBM. König and Loser (2020) compared GLMs, neural networks, and XGBoost for frequency prediction. We want to add to existing research by forecasting pure premium rather than frequency or severity, comparing four methodologies, and analyzing and presenting a wide range of quantitative and qualitative performance evaluation metrics. This paper compares four approaches for predicting pure premium: GLM, AGLM, XGBoost, and neural networks. On a test set, the study compares each model's pros and cons in P&C insurance pricing, as well as its quantitative and qualitative performance. Despite these obstacles, machine learning in actuarial pricing has clear benefits. By may improve risk profiles, pricing accuracy, profitability, and market competitiveness by using massive volumes of data and advanced analytics. However, getting these results needs a holistic approach that balances machine learning's technical sophistication with insurance's regulatory and operational limits. Further research should construct interpretable machine learning models customized to insurance pricing needs and limits while addressing the ethical, regulatory, and implementation obstacles to their adoption. © 2024 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Comparative Analysis of CNN-based Deep Learning Approaches on Complex Activity Recognition 7

Comparative Analysis of CNN-based Deep Learning Approaches o...

引用

作者： Mekruksavanich, Sakorn Jantawong, Ponnipa Jitpattanakul, Anuchit University of Phayao School of Information and Communication Technology Department of Computer Engineering Phayao Thailand King Mongkut's University of Technology North Bangkok Intelligent and Nonlinear Dynamic Innovations Research Center Faculty of Applied Science Department of Mathematics Bangkok Thailand

ISBN: (纸本)9781665495103

For those with Parkinson's disease, freezing of gait (FOG) could be a distressing condition. In individuals with diabetes, FOG could have a significant influence on their quality of life and cause falls, which can be dangerous. A questionnaire has traditionally been used to measure the severity of FOG symptoms;nevertheless, this technique is subjective and can not accurately reflect the intensity of this illness. The adoption of sensor-based devices may give reliable and objective information to follow the progression of symptoms, allowing for better Parkinson's disease management and therapy. A compact deep convolutional neural network comprising squeeze and excitation components, which we called SE-DeepConV model, was designed to identify FOG utilizing wearable data from the sensors. The SE-DeepConV model was constructed and evaluated using a publicly accessible standard FOG dataset, particularly the Daphnet dataset. The SE-DeepConV model outperforms other benchmark deep learning models in terms of accuracy, with the most outstanding accuracy of 95.65%, according to research observations. © 2022 IEEE.

关键词： Deep neural networks

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：