检索结果-内蒙古大学图书馆

Proceedings of the 41st International Conference on Machine Learning

作者： Soo Yong Lee Sunwoo Kim Fanchen Bu Jaemin Yoo Jiliang Tang Kijung Shin Kim Jaechul Graduate School of Artificial Intelligence KAIST Daejeon Republic of Korea Kim Jaechul Graduate School of Artificial Intelligence KAIST Daejeon Republic of Korea and School of Electrical Engineering KAIST Daejeon Republic of Korea School of Electrical Engineering KAIST Daejeon Republic of Korea Department of Computer Science and Engineering Michigan State University Michigan

How would randomly shuffling feature vectors among nodes from the same class affect graph neural networks (GNNs)? The feature shuffle, intuitively, perturbs the dependence between graph topology and features (A-X dependence) for GNNs to learn from. Surprisingly, we observe a consistent and significant improvement in GNN performance following the feature shuffle. Having overlooked the impact of A-X dependence on GNNs, the prior literature does not provide a satisfactory understanding of the phenomenon. Thus, we raise two research questions. First, how should A-X dependence be measured, while controlling for potential confounds? Second, how does A-X dependence affect GNNs? In response, we (i) propose a principled measure for A-X dependence, (ii) design a random graph model that controls A-X dependence, (iii) establish a theory on how A-X dependence relates to graph convolution, and (iv) present empirical analysis on real-world graphs that align with the theory. We conclude that A-X dependence mediates the effect of graph convolution, such that smaller dependence improves GNN-based node classification.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Challenges and Emerging Issues for Generative AI and Hyper Intelligence

Challenges and Emerging Issues for Generative AI and Hyper I...

引用

Cyber Science and Technology Congress (CyberSciTech), IEEE

作者： Jianhua Ma Qun Jin Hui-Huang Hsu John Paul C. Vergara Antonio Guerrieri Claudio Miceli Ao Guo Faculty of Computer and Information Sciences Hosei University Japan Department of Human Informatics and Cognitive Sciences Waseda University Japan Department of Computer Science and Information Engineering Tamkang University Taiwan Department of Information Systems and Computer Science Ateneo de Manila University Philippines ICAR-CNR Italy Graduate Program in Systems Engineering and Computer Science Federal University of Rio de Janeiro Brazil Graduate School of Informatics Nagoya University Japan

ISBN: (数字)9798331532093

ISBN: (纸本)9798331532109

Generative Artificial Intelligence (GenAI) represents a significant milestone in the development of artificial intelligence, bringing sophisticated AI capabilities into daily life and work. As we approach the era of Hyper Intelligence (Hyper-I), a variety of critical challenges and emerging issues have come to light, ranging from computational complexity to ethical concerns. This paper explores the evolution of AI from the perspective of human learning, comparing machine and human intelligence, and identifying key considerations for the development of future AI systems. It also highlights the growing importance of regulating advanced AI models, such as Reinforcement Learning-based Long-Term Planning Agents, to ensure that Hyper-I remains under human control. Additionally, the paper discusses the computational complexity of transformer-based models, their applicability to intractable problems, and their role in cognitive building systems and resource-constrained environments through TinyML. By analyzing these pressing challenges, this work provides insights into the future of AI and the path toward responsible innovation in generative and hyper-intelligent systems.

关键词： Hyper-intelligent systems Ethics Technological innovation Generative AI Computational modeling Tiny machine learning Pressing Transformers Regulation Computational complexity

来源：评论

学校读者我要写书评

暂无评论

Extended Mixed Filtering based on Zonotopic and Gaussian Uncertainties for Discrete-Time Nonlinear Systems

Extended Mixed Filtering based on Zonotopic and Gaussian Unc...

引用

European Control Conference (ECC)

作者： Alesi A. De Paula Guilherme V. Raffo Bruno O. S. Teixeira Graduate Program in Electrical Engineering - Universidade Federal de Minas Gerais (UFMG) Belo Horizonte MG Brazil Department of Electronic Engineering Universidade Federal de Minas Gerais (UFMG) Belo Horizonte MG Brazil

ISBN: (数字)9783907144107

ISBN: (纸本)9798331540920

In this paper, we propose a mixed filter for discrete- time nonlinear dynamical systems whose uncertainties are composed of both deterministic and stochastic terms. In practice, such mixed composition of uncertainties may appear when assimilating measured data and approximating models. The unknown-but-bounded terms are here represented by zonotopes, which are efficient representations of centrally symmetric convex polytopes. In turn, the stochastic terms are represented by Gaussian random vectors (GRVs) which address confidence regions with high probability. The proposed state estimator is based on linearized models, using the quasi-linear parameter- varying (LPV) approach. The effectiveness of our proposal is illustrated in two case studies.

关键词： Uncertainty Filtering Measurement uncertainty Stochastic processes Europe Vectors Mathematical models

来源：评论

学校读者我要写书评

暂无评论

Leader-Follower UGV-UAV Formation as Control Paradigm for Package Delivery

Leader-Follower UGV-UAV Formation as Control Paradigm for Pa...

引用

International Conference on Unmanned Aircraft Systems (ICUAS)

作者： Viní cius Pacheco Bacheti Alexandre Santos Brandã o Má rio Sarcinelli-Filho Graduate Program on Electrical Engineering Federal University of Esp&#x00ED rito Santo Brazil Department of Electrical Engineering Federal University of Vi&#x00E7 osa Brazil

ISBN: (数字)9781665405935

ISBN: (纸本)9781665405942

This paper addresses the problem of package delivery by a drone taking off of a ground vehicle, flying towards a desired delivery point and flying back to land on the ground vehicle. Specifically, the final step, when the drone approaches the ground vehicle and lands on it is dealt with here. To do that a ground mobile robot, a Pioneer 3-DX unicycle platform, is used to play the role of the delivery truck, whereas the drone is the Bebop 2 quadrotor. To control such a formation the leader-follower control paradigm is adopted, in which the Pioneer 3-DX is the leader and the quadrotor is the follower, which mimics the movement of the leader. According to such a control paradigm, the two vehicles are independently controlled. A first controller is responsible for the leader and a second one is responsible for the follower. Aiming at a package delivery application, the leader robot is programmed to follow a given path, thus playing the role of a delivery truck, whereas a trajectory-tracking controller is designed for the follower, having the leader as the trajectory to be tracked. Therefore, the leader-follower control paradigm fits quite well the application in mind, which is the reason for this proposal. Simulation results considering such a system are presented here, which validate the proposed approach.

关键词： Simulation MIMICs Autonomous aerial vehicles Land vehicles Trajectory Proposals Mobile robots

来源：评论

学校读者我要写书评

暂无评论

Correcting Negative Bias in Large Language Models through Negative Attention Score Alignment

arXiv

引用

arXiv 2024年

作者： Yu, Sangwon Song, Jongyoon Hwang, Bongkyu Kang, Hoyoung Cho, Sooah Choi, Junhwa Joe, Seongho Lee, Taehee Gwon, Youngjune L. Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of Samsung SDS Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of

A binary decision task, like yes-no questions or answer verification, reflects a significant real-world scenario such as where users look for confirmation about the correctness of their decisions on specific issues. In this work, we observe that language models exhibit a negative bias in the binary decisions of complex reasoning tasks. Based on our observations and the rationale about attention-based model dynamics, we propose a negative attention score (NAS) to systematically and quantitatively formulate negative bias. Based on NAS, we identify attention heads that attend to negative tokens provided in the instructions as answer candidate of binary decisions, regardless of the question in the prompt, and validate their association with the negative bias. Additionally, we propose the negative attention score alignment (NASA) method, which is a parameter-efficient fine-tuning technique to address the extracted negatively biased attention heads. Experimental results from various domains of reasoning tasks and large model search space demonstrate that NASA significantly reduces the gap between precision and recall caused by negative bias while preserving their generalization abilities. Our codes are available at https://***/ysw1021/NASA. 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

Robust System Reconfiguration in the Presence of Uncertain Loads and Renewable Generation

Robust System Reconfiguration in the Presence of Uncertain L...

引用

IEEE Texas Power and Energy Conference (TPEC)

作者： Meisam Mahdavi Konrad Schmitt Stephen Bayne Manohar Chamana Department of Electrical Engineering University of Jaén Linares Jaén Spain Electrical & Computer Engineering Department Texas Tech University Lubbock TX USA Renewable Energy Program Texas Tech University Lubbock USA

Although distributed generation (DG) operation in modern distribution systems reduces energy losses, DG penetration in such networks is restricted by technical and investment limitations. In order to reduce these losses more efficiently, active distribution systems should be reconfigured based on their time-varying and uncertain load and renewable generation levels. However, considering demand variability in distribution systems’ reconfiguration increases the mathematical efforts, computational burden, and processing time. Besides, the reconfiguration model must be robust enough to avoid consecutive reconfigurations under small load and generation variations. Accordingly, reconfiguration strategies should be chosen so that they have enough robustness and good convergence speed regarding load and generation uncertainty and variability. Therefore, the present paper develops an effective and robust strategy for active distribution systems’ reconfiguration with load and renewable generation variability and uncertainty. The model is submitted to two standard distribution systems where the results are compared with other models presented by the literature. The comparison highlights the robustness and computational time efficiency of the presented model.

关键词： Renewable energy sources Energy loss Uncertainty Computational modeling Distribution networks Robustness Topology

来源：评论

学校读者我要写书评

暂无评论

Sample Complexity of Kernel-Based Q-Learning

arXiv

引用

arXiv 2023年

作者： Yeh, Sing-Yuan Chang, Fu-Chieh Yueh, Chang-Wei Wu, Pei-Yuan Bernacchia, Alberto Vakili, Sattar MediaTek Research Graduate Program of Data Science National Taiwan University Academia Sinica Taiwan Graduate Institute of Communication Engineering National Taiwan University Taiwan Department of Electrical Engineering National Taiwan University Taiwan

Modern reinforcement learning (RL) often faces an enormous state-action space. Existing analytical results are typically for settings with a small number of state-actions, or simple models such as linearly modeled Q-functions. To derive statistically efficient RL policies handling large state-action spaces, with more general Q-functions, some recent works have considered nonlinear function approximation using kernel ridge regression. In this work, we derive sample complexities for kernel based Q-learning when a generative model exists. We propose a nonparametric Q-learning algorithm which finds an ǫ-optimal policy in an arbitrarily large scale discounted MDP. The sample complexity of the proposed algorithm is order optimal with respect to ǫ and the complexity of the kernel (in terms of its information gain). To the best of our knowledge, this is the first result showing a finite sample complexity under such a general model. © 2023, CC BY.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

The Smart Analysis of Performing Scalable Inference for Big Data Analytics

The Smart Analysis of Performing Scalable Inference for Big ...

引用

2023 International Conference on Power Energy, Environment and Intelligent Control, PEEIC 2023

作者： Bhandarkar, Vishakha D. Rongali, Aruna Sri Shamout, Mohamed Dawood Pund, Sachin Nama, Gigih Forda Patil, Shrishailappa Yeshwantrao Chavan College of Engineering Applied Mathematics and Humanities India University of the Cumberlands Graduate Doctoral Information Technology Information Technology Emphasis/Information Technology United States College of Business Administration University of Sharjah Department of Management Sharjah United Arab Emirates Shri Ramdeobaba College of Engineering and Management Department of Industrial Engineering Maharashtra Nagpur India Doctoral Program of Environtmental Science University of Lampung Department of Informatics Bandar Lampung Indonesia Vishwakarma Institute of Technology Computer Engineering Pune India

ISBN: (纸本)9798350357769

This paper examines the reproducibility of massive information analytics under particular factors. The paper proposes the 'performing Scalable Inference' technique to cope with scalability troubles and to exploit current big statistics platforms for efficient computing and statistics garage of the statistics. In particular, the paper describes how to perform leak-free, parallelizable visible analytics over massive datasets using present extensive records analytics frameworks such as Apache Flink. This method presents an automated manner to execute analytics that preserves reproducibility and the ability to make adjustments without re-running the entire technique. The paper also demonstrates how these analytics may help several real-world use instances, explore affected person cohorts for studies, and develop stratified patient cohorts for hospital therapy. In the end, the paper observes how the proposed method may be exercised within the real world. Actively scalable inference for massive information analytics is pivotal in optimizing decision-making and allocation of assets. Typically, such inferences are made based on information accumulated from numerous sources, databases, unstructured data, and different digital sources. So one can ensure scalability, a complete cloud-primarily based platform has to be hired. This solution will permit the ***, deploying the essential records series and evaluation algorithms are prime here. It could permit the platform to recognize the styles inside the statistics and discover any ability correlations or traits. Additionally, predictive analytics and system mastering strategies may be incorporated to provide insights into the results of the information. In the long run, by leveraging those techniques, the platform can draw efficient inferences and appropriately compare situations in an agile and green way.. © 2023 IEEE.

关键词： adjustments an a lytics automated preserves statistics

来源：评论

学校读者我要写书评

暂无评论

An Efficient Formulation for Optimal Placement of Capacitor Banks in Distribution Feeders Connected to Time-Varying Loads

An Efficient Formulation for Optimal Placement of Capacitor ...

引用

IEEE Kansas Power and Energy Conference

作者： Meisam Mahdavi Augustine Awaafo Konrad Schmitt Manohar Chamana Department of Electrical Engineering University of Jaén Linares Jaén Spain Electrical and Computer Engineering Department Texas Tech University Lubbock TX USA Renewable Energy Program Texas Tech University Lubbock USA

The capacitor placement in radial feeders is an efficient way to mitigate energy losses of distribution networks, where the power converted to useless heat is much higher than transmission system. These considerable power losses increase network operating costs and also decrease quality and quantity of delivered power. Shunt capacitors by compensating part of load reactive power reduce magnitude of feeder current and subsequent distribution losses. However, electricity demand variations with time affects power distributed through feeders and therefore have important role in energy losses determination. Thus, load variability may change the optimal places of capacitors, but considering whole load variations substantially increases the burden and computational time of capacitor placement problem in large-sized distribution systems. Therefore, current study poses an effective capacitor placement formulation for energy loss reduction in the presence of time-varying loads. The numerical results analysis demonstrates the good performance of the presented framework for the optimal placement of capacitor banks in radial distribution feeders with variable loads.

关键词：

来源：评论

学校读者我要写书评

暂无评论

The modified Allen test and a novel objective screening algorithm for hand collateral circulation using differential photoplethysmography for preoperative assessment: a pilot study

引用

Journal of Medical engineering and Technology 2020年第2期44卷 82-93页

作者： Elwali, Ahmed Moussavi, Zahra Biomedical Engineering Program University of Manitoba Winnipeg Canada Electrical and Computer Engineering Department University of Manitoba Winnipeg Canada

Hand-ischemia and artery-thrombosis are serious complications of hand artery cannulation/grafting. Screening the patency of hand arteries reduces the complications. Modified Allen Test (MAT), Barbeau's test, and Doppler/Duplex-ultrasonic test screen hand arterial patency. Each of them, however, either lacks objectivity, requires subject cooperation, or is not readily available in operating rooms. In this pilot study, we aim to provide a novel, objective, reliable, and inexpensive alternative using differential photo-plethysmography (PPG) measurement, and an artery selection algorithm. For 11 participants, MAT was executed, and PPG measurements of the thumb and little fingers were recorded for left and right hands separately. For each hand, four evaluations were conducted, then fed into the artery selection algorithm. Our proposed procedure has shown congruent results with the anatomy of blood circulation. It identified two arteries in two participants causing >50% blood reduction in the hand due to their occlusion. It also showed an elevated risk of accessing an artery in another 8 participants. Once the results of this study are confirmed in larger samples, the proposed procedure/algorithm may provide many indicators of the hand blood circulation, the effect of an artery occlusion on the hand, and a recommendation for selecting the best artery to access with the lowest possible complications. © 2020, © 2020 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Diseases

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：