检索结果-内蒙古大学图书馆

ieee 20th International conference on Automation Science and Engineering (CASE)

作者： Wang, Xizhe Zhao, Hongbo Zhao, Qianchuan Jiang, Benben Tsinghua Univ Ctr Intelligent & Networked Syst Dept Automat Beijing Peoples R China Princeton Univ Dept Chem & Biol Engn Princeton NJ 08544 USA

ISBN: (纸本)9798350358513;9798350358520

The discovery of partial differential equations (PDEs) directly from data has significant implications for automation and manufacturing fields. This approach leverages advanced computational techniques and data-driven methodologies to identify the underlying mathematical models that govern the physical processes in various systems. In this article, a deep symbolic regression framework is developed for PDE discovery which is enhanced by reinforcement learning with spatiotemporal rewards. Within this framework, a dense encoderdecoder network is proposed to adeptly extract the information of temporal dynamics related to PDEs from the data, and a refined reinforcement learning reward shaping technique is further put forward that incorporates the extracted dynamics, specifically designed for PDE symbolic regression tasks. An example of Cahn-Hilliard equation, which is widely used to describe the phase separation process in multi-component mixture systems, is used to validate the effectiveness of the proposed approach. The experimental results show that the proposed deep symbolic regression approach outperforms the baselines in identifying PDEs, achieving higher accuracy with less data. Additionally, we provide a thorough analysis on how adjustable parameters affect the PDE discovery performance, further verifying the effectiveness of the proposed approach.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Resilient Machine learning (rML) Against Adversarial Attacks on Industrial control systems 20

Resilient Machine Learning (rML) Against Adversarial Attacks...

引用

20th ACS/ieee International conference on Computer systems and Applications (AICCSA)

作者： Yao, Likai Shao, Sicong Hariri, Salim Univ Arizona Dept Elect & Comp Engn Tucson AZ 85721 USA Univ North Dakota Sch Elect Engn & Comp Sci Grand Forks ND 58202 USA

ISBN: (纸本)9798350319439

Machine learning (ML) algorithms have been widely used in many critical automated systems, including as a technique in Dynamic data driven Applications systems (DDDAS)-based methods and areas such as financial trading, autonomous vehicles, and intrusion detection systems. However, malicious adversaries have strong interests in manipulating the operations of machine learning algorithms to achieve their objectives of gaining financial, social, or political influence. Adversarial ML (AML) users can be classified based on their capabilities and goals into three types: Adversary who has full knowledge of the ML models and parameters (white-box scenario), partial knowledge of ML models (gray-box scenario), and one who does not have any knowledge and uses guessing techniques to figure out the ML model and its parameters (black-box scenario). In these scenarios, the adversaries attempt to maliciously manipulate the model/data either during training or testing. Defending against these AML attacks can be successful by following methods such as making the ML model robust, validating and verifying inputs and outputs, and changing the ML architecture. This paper presents a resilient machine learning (rML) against adversarial attacks by dynamically conducting feature space anonymization and model randomization in ML services such that the adversaries lack knowledge about the feature space and model used and consequently prevent them from maliciously manipulating the ML operations during the runtime. In our approach, the rML utilizes autoencoders as an anonymization technique for encoding feature space to minimize the effect of adversarial samples. The rML method is evaluated using the benchmarking Industrial control systems (ICS) data and the corresponding adversarial data generated using the Jacobian-based Saliency Map Attack (JSMA) method. The experiment demonstrated that the proposed approach can detect attacks targeting ICS and prevent adversarial attacks compromising ML mode

关键词： Dynamic data driven Application systems resiliency adversarial machine learning moving target defense resilient decision support Anonymization

来源：评论

学校读者我要写书评

暂无评论

Redefining Urban Traffic Dynamics With TCN-FL driven Traffic Prediction and control Strategies

引用

ieee ACCESS 2024年 12卷 115386-115399页

作者： Raghunath, K. M. Karthick Bhat, C. Rohith Kumar, Venkatesan Vinoth Kannan, Velmurugan Athiyoor Mahesh, T. R. Manikandan, K. Krishnamoorthy, N. JAIN Deemed Be Univ Dept Comp Sci Global Campus Bengaluru 560034 Karnataka India Saveetha Univ Saveetha Inst Med & Tech Sci Saveetha Sch Engn Engn Chennai 602105 India VIT Sch Comp Sci Engn & Informat Syst SCORE Vellore 632014 Tamil Nadu India Koneru Lakshmaiah Educ Fdn Dept Comp Sci & Engn Guntur 522302 India JAIN Deemed Be Univ Sch Comp Sci & Engn Bengaluru 560034 Karnataka India VIT Sch Comp Sci & Engn SCOPE Vellore 632014 Tamil Nadu India

The smart city traffic management domain is perpetually a crucial sector that requires innovative strategies due to expanding urbanization and vehicle use. In this study, we have introduced a traffic prediction and handling system that utilizes Temporal Convolutional Networks (TCNs) combined with Federated learning (FL) to deal with urban traffic effectively. This approach leverages the sophisticated functionalities of TCNs to evaluate and estimate traffic trends, such as congested phases, traffic flow, and ideal mobility routes. The system guarantees data privacy and utilizes decentralized information analysis using Federated learning. In this approach, every point in the intelligent city network, including traffic sensors and cameras, serves to collectively comprehend traffic patterns without disclosing raw data. Using this cooperative method not only improves the model's ability to forecast outcomes accurately but also enables efficient real-time traffic management, with the ability to adapt to changing conditions. The method has shown significant efficacy in enhancing traffic flow and mitigating congestion. The critical criteria are a 20% drop in average commuting times, a 25% drop in traffic congestion, and a 15% enhancement in emergency response times. These statistics highlight the system's efficiency in improving urban traffic control by integrating modern technologies.

关键词： Vehicle dynamics Urban areas Real-time systems Traffic control data models Optimization Predictive models Decentralized applications Decentralized learning mobile patterns transport traffic prediction accuracy normalization temporal convolution

来源：评论

学校读者我要写书评

暂无评论

data-driven Structure Identification of Takagi Sugeno Fuzzy Models Using a Bounded Error Approach

Data-Driven Structure Identification of Takagi Sugeno Fuzzy ...

引用

ieee International conference on Fuzzy systems (FUZZ-ieee)

作者： Wittich, Felix Kroll, Andreas Univ Kassel Dept Measurement & Control Kassel Germany

ISBN: (纸本)9798350332285

Takagi Sugeno fuzzy models are a widely used model class for data driven identification of nonlinear systems. Besides good approximation quality, the local model approach allows for good interpretability and the utilization of linear control strategies in dynamic models. However a major challenge is the identification of the model structure, i.e. determining the number of local models and the partitioning parameters. A new method for identification of Takagi Sugeno fuzzy models inspired by bounded error methods is presented. The method finds partitions of the data that are defined by bounding linear inequalities and therefore identifies locally linear system behavior. The error bound as a tuning parameter is used to adjust the compromise between model accuracy and sparsity. The presentation of the method is complemented by two case studies.

关键词： Fuzzy Modeling Bounded Error Structure Identification Partitioning data-driven Modeling

来源：评论

学校读者我要写书评

暂无评论

Analysis of factors affecting the future of the Game industry through Machine learning 3

Analysis of factors affecting the future of the Game industr...

引用

3rd International conference on Communication, control, and Intelligent systems, CCIS 2024

作者： Junaid, Mohammed Taneja, Mehak Sehgal, Swastik Bhagwan Parshuram Institute of Technology Ggsipu Dept. of Cse Delhi India Indira Gandhi Delhi Technical University for Women Dept. of Cse & Ai Delhi India Dps R K Puram Delhi India

ISBN: (纸本)9798331528201

The video game industry is a significant economic force, generating billions of dollars in revenue annually. This paper analyses the factors influencing consumer preferences for different types of video games, focusing on platform availability, pricing, game genres, and evolving market trends. data was collected through a survey of over 250 respondents, capturing demographic attributes and gaming preferences. Machine learning techniques, including Decision Tree, Random Forest, and LightGBM, were applied to predict platform preferences and recommend game genres based on age. The Decision Tree model outperformed others with an accuracy of 99.98%, demonstrating its suitability for both tasks. This study provides actionable insights into consumer behaviour, emphasizing the role of brand reputation, marketing efforts, and franchise value in shaping game success. These findings aim to inform game developers and publishers in crafting personalized, data-driven strategies to enhance player engagement and satisfaction. © 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

Online Reinforcement learning control by Direct Heuristic Dynamic Programming: From Time-driven to Event-driven

引用

ieee TRANSACTIONS ON NEURAL NETWORKS AND learning systems 2022年第8期33卷 4139-4144页

作者： Zhao, Qingtao Si, Jennie Sun, Jian Beijing Inst Technol Key Lab Intelligent Control & Decis Complex Syst Beijing 100081 Peoples R China Arizona State Univ Dept Elect Comp & Energy Engn Tempe AZ 85287 USA

In this work, time-driven learning refers to the machine learning method that updates parameters in a prediction model continuously as new data arrives. Among existing approximate dynamic programming (ADP) and reinforcement learning (RL) algorithms, the direct heuristic dynamic programming (dHDP) has been shown an effective tool as demonstrated in solving several complex learning control problems. It continuously updates the control policy and the critic as system states continuously evolve. It is therefore desirable to prevent the time-driven dHDP from updating due to insignificant system event such as noise. Toward this goal, we propose a new event-driven dHDP. By constructing a Lyapunov function candidate, we prove the uniformly ultimately boundedness (UUB) of the system states and the weights in the critic and the control policy networks. Consequently, we show the approximate control and cost-to-go function approaching Bellman optimality within a finite bound. We also illustrate how the event-driven dHDP algorithm works in comparison to the original time-driven dHDP.

关键词： Neural networks Task analysis Convergence Stability criteria Heuristic algorithms Approximation algorithms learning systems Direct heuristic dynamic programming (dHDP) event-driven time-driven dHDP reinforcement learning (RL)

来源：评论

学校读者我要写书评

暂无评论

Distributed Economic Dispatch for Multi-area Power systems Based on Multi-objective Differential Evolution Algorithm 13

Distributed Economic Dispatch for Multi-area Power Systems B...

引用

13th ieee data driven control and learning systems conference, DDCLS 2024

作者： Ren, Xuyang Bu, Xuhui Yin, Yanling Li, Xuan Zhang, Yan School of Electrical Engineering and Automation Henan Polytechnic University Jiaozuo454003 China School of Business Administration Henan Polytechnic University Jiaozuo454003 China College of Business Zhengzhou Technology and Business University Zhengzhou451400 China

ISBN: (纸本)9798350361674

Economic dispatch of a multi-area interconnected power system with wind, solar, and energy storage units is a typical non-convex and nonlinear optimization problem, which is difficult to be solved by the existing mathematical programming methods. In this paper, a distributed multi-objective differential evolution algorithm is proposed. Firstly, to improve the global search ability of the multi-objective differential evolutionary algorithm, the population is partitioned using a competitive mechanism. A multi-strategy scheme is used according to the partition to enhance the diversity of the algorithm. Secondly, the ieee 39-bus system is decomposed into several sub-areas using the bus tearing method and the interval information is transmitted through the boundary tie-lines. Then a distributed method is proposed to be applied to the above algorithm, where each sub-algorithm solves the sub-problem associated with each area. Finally, the ieee 39-bus system is divided into two-area and three-area interconnected systems, and the performance of the algorithm is verified on the modified ieee 39-bus system. The experimental results verify the effectiveness of the improved multi-objective differential evolution optimization algorithm and the feasibility of the distributed approach. © 2024 ieee.

关键词： Optimization algorithms

来源：评论

学校读者我要写书评

暂无评论

Intelligent Network Slicing for B5G and 6G: Resource Allocation, Service Provisioning, and Security

引用

ieee WIRELESS COMMUNICATIONS 2024年第3期31卷 271-277页

作者： Wang, Jiadai Li, Yuanhao Liu, Jiajia Kato, Nei Northwestern Polytech Univ China Xian Peoples R China Tohoku Univ Tohoku Japan

B5G and 6G are designed to achieve break-throughs in mobile networks, and are expected to have features such as seamless global coverage, full virtualization, and ubiquitous intelligence. These promising features will spawn lots of new verticals and applications, creating an urgent need for on-demand networks. At this point, network slicing has been recognized as an excellent solution to this challenge due to its advantages of flexible customization and logical isolation. driven by massive data, intelligent network slicing is further endowed with capabilities of real-time perception, accurate prediction, and adaptive decision making, which brings unlimited potential for the development of B5G/6G. To this end, we investigate the application and prospect of intelligent slicing for B5G/6G in three areas, that is, resource allocation, service provisioning, and security. We also list the specific problems in these areas according to each slice lifecycle phase. In addition, a case study of using reinforcement learning for slice admission control is provided to improve the slice admission rate under limited resources and illustrate the advantages of intelligent network slicing.

关键词： 6G mobile communication Intelligent networks Network slicing Admission control Decision making Reinforcement learning Real-time systems

来源：评论

学校读者我要写书评

暂无评论

Detecting and Adapting to CSI Feedback Drift in data-driven Compression for MIMO systems 1

Detecting and Adapting to CSI Feedback Drift in Data-Driven ...

引用

1st ieee International conference on Machine learning for Communication and Networking (ieee ICMLCN)

作者： Wang, Xinyi Mestav, Kursat Rasim Nuzman, Carl Saniee, Iraj Cornell Univ Elect & Comp Engn 116 Hoy Rd Ithaca NY 14853 USA Nokia Bell Labs AI Lab 600 Mt Ave Murray Hill NJ 07974 USA

ISBN: (纸本)9798350343205;9798350343199

Deep neural networks make it possible to learn key characteristics of data without having to assume mathematically tractable models. This in turn results in an ability to compress data in a model-free way. One of the promising areas for the application of deep learning to the physical layer of communication networks is compression. Hundred-fold compression with a small loss of the CSI in massive MIMO systems has been shown to be both feasible and necessary. However, model-free and data-driven compression comes with a downside: the encoding and decoding models need to be trained on a large set of CSI arrays indicative of a wide spectrum of propagation and environmental conditions. As a result, in the early stages of the deployment of deep CSI compression models, it would be necessary to detect if and when users' channels have drifted significantly away from the distribution of the CSI data on which the deep compression model was trained. In this paper, we present both 1) a technique for detecting harmful channel drift and 2) a lightweight scheme for fine-tuning the deep compression models to adjust to such shifts. Using public-domain synthetic channel data as well as 3GPP-compliant simulated data, we demonstrate the practicality of our proposed deep compression and detection framework. We close with recommendations for a viable implementation of the proposed drift detection by the standards bodies.

关键词： channel state information massive MIMO deep neural network compression data distribution distribution drift detection

来源：评论

学校读者我要写书评

暂无评论

Pupil Diameter Classification using Machine learning During Human-Computer Interaction

Pupil Diameter Classification using Machine Learning During ...

引用

ieee International conference on Omni-Layer Intelligent systems (ieee COINS)

作者： Azizinezhad, Parastoo Ghonchi, Hamidreza Chowdhury, Anirban Univ Essex Colchester Essex England

ISBN: (纸本)9798350349603;9798350349597

The significance of eye tracking and pupillary responses spans various disciplines, especially for people with severe disabilities. Effortful decision-making is marked by pupil dilation, reflecting increased cognitive load and can be used as a potential measure for system adaptation to users' mental states. This study investigates the classification of pupil diameter data to differentiate between decision-making and focus time in a mobile robot navigation task. data were collected from 19 healthy participants utilizing an eye-tracking-based user interface to control the robot's movements along pre-set paths. This paper presents a deep learning and SVM-based classification approach to distinguish focus and decision-making from pupil diameter patterns, offering insights for achieving an adaptive command selection approach. On average, the proposed deep learning model has an average accuracy of over 82% in classifying the data for participants using pupil diameter data.

关键词： Eye-Tracking Pupil diameter Human-Computer interaction

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：