检索结果-内蒙古大学图书馆

online learning algorithms for Double-Weighted Least Squares Twin Bounded Support Vector Machines

NEURAL PROCESSING LETTERS 2017年第1期45卷 319-339页

作者： Li, Juntao Cao, Yimin Wang, Yadi Xiao, Huimin Henan Normal Univ Sch Math & Informat Sci Henan Engn Lab Big Data Stat Anal & Optimal Contr Xinxiang 453007 Peoples R China Henan Univ Econ & Law Coll Comp & Informat Engn Zhengzhou 450002 Peoples R China

Twin support vector machine with two nonparallel classifying hyperplanes and its extensions have attracted much attention in machine learning and data mining. However, the prediction accuracy may be highly influenced when noise is involved. In particular, for the least squares case, the intractable computational burden may be incurred for large scale data. To address the above problems, we propose the double-weighted least squares twin bounded support vector machines and develop the online learning algorithms. By introducing the double-weighted mechanism, the linear and nonlinear double-weighted learning models are proposed to reduce the influence of noise. The online learning algorithms for solving the two models are developed, which can avoid computing the inverse of the large scale matrices. Furthermore, a new pruning mechanism which can avoid updating the kernel matrices in every iteration step for solving nonlinear model is also developed. Simulation results on three UCI data with noise demonstrate that the online learning algorithm for the linear double-weighted learning model can get least computation time as well considerable classification accuracy. Simulation results on UCI data and two-moons data with noise demonstrate that the nonlinear double-weighted learning model can be effectively solved by the online learning algorithm with the pruning mechanism.

关键词： Support vector machine Twin bounded support vector machine Double-weighted mechanism online learning algorithms Pruning mechanism

来源：评论

学校读者我要写书评

暂无评论

online learning Approach Based on Recursive Formulation for Twin Support Vector Machine and Sparse Pinball Twin Support Vector Machine

引用

NEURAL PROCESSING LETTERS 2023年第4期55卷 5143-5165页

作者： Shadiani, Abolfazl Hasanzadeh Shoorehdeli, Mahdi Aliyari KN Toosi Univ Technol Fac Elect Engn Tehran Tehran Iran

In this paper, an online approach was proposed for twin support vector machine motivated by online learning algorithms for double-weighted least squares twin bounded support vector machines. In many applications for training, data are available online, and batch training methods are not suitable because of space and time requirements. For the online method proposed in this paper, the online learning method was created by recursive relation of twin support vector machine in two linear and nonlinear cases, which avoids calculating inverse matrices in every repetition step. Thus, only the inverse matrix in the initial step must be calculated, and every repetition step is calculated recursively from the previous step, which causes the training time to decrease without losing accuracy. Moreover, for studying the effectiveness of the proposed approach, this online approach was used for sparse pinball twin support vector machine, and simulation results indicated this online approach not only did not reduce accuracy but also, for some datasets, increased accuracy for online cases.

关键词： Support vector machine Twin support vector machine Sparse pinball twin support vector machine online learning algorithms Quadratic programming classification

来源：评论

学校读者我要写书评

暂无评论

HierFedML: Aggregator Placement and UE Assignment for Hierarchical Federated learning in Mobile Edge Computing

引用

IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS 2023年第1期34卷 328-345页

作者： Xu, Zichuan Zhao, Dapeng Liang, Weifa Rana, Omer F. Zhou, Pan Li, Mingchu Xu, Wenzheng Li, Hao Xia, Qiufen Dalian Univ Technol Sch Software Key Lab Ubiquitous Network & Serv Software Liaonin Dalian 116621 Liaoning Peoples R China City Univ Hong Kong Dept Comp Sci Hong Kong Peoples R China Cardiff Univ Phys Sci & Engn Coll Cardiff CF10 3AT Wales Huazhong Univ Sci & Technol Hubei Engn Res Ctr Big Data Secur Sch Cyber Sci & Engn Hubei Key Lab Distributed Syst Secur Wuhan 430074 Hubei Peoples R China Sichuan Univ Dept Comp Sci Chengdu 610065 Sichuan Peoples R China Ningxia Res Inst China Coal Res Inst Beijing 100013 Peoples R China Dalian Univ Technol Int Sch Informat Sci & Engn Key Lab Ubiquitous Network & Serv Software Liaonin Dalian 116621 Liaoning Peoples R China

Federated learning (FL) is a distributed machine learning technique that enables model development on user equipments (UEs) locally, without violating their data privacy requirements. Conventional FL adopts a single parameter server to aggregate local models from UEs, and can suffer from efficiency and reliability issues - especially when multiple users issue concurrent FL requests. Hierarchical FL consisting of a master aggregator and multiple worker aggregators to collectively combine trained local models from UEs is emerging as a solution to efficient and reliable FL. The placement of worker aggregators and assignment of UEs to worker aggregators plays a vital role in minimizing the cost of implementing FL requests in a Mobile Edge Computing (MEC) network. Cost minimization associated with joint worker aggregator placement and UE assignment problem in an MEC network is investigated in this work. An optimization framework for FL and an approximation algorithm with an approximation ratio for a single FL request is proposed. online worker aggregator placements and UE assignments for dynamic FL request admissions with uncertain neural network models, where FL requests arrive one by one without the knowledge of future arrivals, is also investigated by proposing an online learning algorithm with a bounded regret. The performance of the proposed algorithms is evaluated using both simulations and experiments in a real testbed with its hardware consisting of server edge servers and devices and software built upon an open source hierarchical FedML (HierFedML) environment. Simulation results show that the performance of the proposed algorithms outperform their benchmark counterparts, by reducing the implementation cost by at least 15% per FL request. Experimental results in the testbed demonstrate the performance gain using the proposed algorithms using real datasets for image identification and text recognition applications.

关键词： Mobile edge computing federated learning aggregator placement and UE assignment hierarchical federated learning framework approximation algorithms online learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Supervised learning in Model Reference Adaptive Sliding Mode Control

引用

INTERNATIONAL JOURNAL OF CONTROL AUTOMATION AND SYSTEMS 2024年第6期22卷 2007-2015页

作者： Makke, Omar Lin, Feng Wayne State Univ Dept Elect & Comp Engn Detroit MI 48202 USA

The well known back-propagation algorithm has revolutionized machine learning and artificial intelligence, particularly in neural network applications. Although gradient descent-based algorithms are utilized in control applications, they are not as prevalent as in neural network applications. This discrepancy can be attributed to the successful development of various adaptation laws which ensure system stability while meeting the required design criteria. Many of these laws can be found in model reference adaptive control (MRAC) and adaptive sliding mode control (ASMC). This paper investigates the applicability of the Brandt-Lin (B-L) learning algorithm, mathematically equivalent to the back-propagation algorithm, in adaptive control applications. We find that combining the B-L learning algorithm with SMC yields a robust controller suitable for model reference adaptive sliding mode control (MRA-SMC). The controller is applicable to linear and a class of nonlinear dynamic systems and is suitable for efficient implementation. We derive the stability criteria for this controller and conduct simulations to study the adaptation's impact on chattering. Our work exemplifies one approach to adopt the back-propagation algorithm in control applications.

关键词： Adaptive control adaptive sliding mode control (ASMC) back-propagation algorithm chattering reduction model reference adaptive control (MRAC) online learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Actor-critic reinforcement learning for the feedback control of a swinging chain

引用

IFAC-Papersonline 2018年第13期51卷 378-383页

作者： Dengler, C. Lohmann, B. Technical University of Munich Germany

Reinforcement learning offers a multitude of algorithms allowing to learn a nonlinear controller by interacting with the system without the need for a model of the plant. In this paper we investigate the suitability of online learning algorithms for a control task with incomplete state information. The system under consideration is a swinging chain that needs to be stabilized at a desired position, a problem that is occurring e.g. with bridge cranes with each change in the crane position. The measurable states are the position, velocity, angle and angular velocity at the top of the chain. A solution of the control problem based on an approximation of the chain as a continuous cable exists in the literature, see d’ Andrea-Novel and Coron (2000), which is included in the comparison as a reference for the control performance of the learned controllers. © 2018

关键词： learning algorithms Approximation algorithms Chains Controllers Cranes Dynamic models Feedback Reinforcement learning Actor critic reinforcement learning Control performance Desired position Function approximation Non linear control Non linear controllers online learning algorithms State information

来源：评论

学校读者我要写书评

暂无评论

Optimal learning algorithms for Stochastic Inventory Systems with Random Capacities

引用

PRODUCTION AND OPERATIONS MANAGEMENT 2020年第7期29卷 1624-1649页

作者： Chen, Weidong Shi, Cong Duenyas, Izak Univ Michigan Ind & Operat Engn Ann Arbor MI 48109 USA Univ Michigan Ross Sch Business Technol & Operat Ann Arbor MI 48109 USA

We propose the first learning algorithm for single-product, periodic-review, backlogging inventory systems with random production capacity. Different than the existing literature on this class of problems, we assume that the firm has neither prior information about the demand distribution nor the capacity distribution, and only has access to past demand and supply realizations. The supply realizations are censored capacity realizations in periods where the policy need not produce full capacity to reach its target inventory levels. If both the demand and capacity distributions were known at the beginning of the planning horizon, the well-known target interval policies would be optimal, and the corresponding optimal cost is referred to as the clairvoyant optimal cost. When such distributional information is not available a priori to the firm, we propose a cyclic stochastic gradient descent type of algorithm whose running average cost asymptotically converges to the clairvoyant optimal cost. We prove that the rate of convergence guarantee of our algorithm is O(1/T), which is provably tight for this class of problems. We also conduct numerical experiments to demonstrate the effectiveness of our proposed algorithms.

关键词： inventory random capacity online learning algorithms regret analysis

来源：评论

学校读者我要写书评

暂无评论

Mobility Management in Low Altitude Heterogeneous Networks Using Reinforcement learning Algorithm 1

引用

5th International Conference on Space Information Networks (SINC)

作者： Hou, Yunpeng Wang, Chao He, Huasen Yang, Jian Univ Sci & Technol China Hefei Peoples R China

ISBN: (数字)9789811619670

ISBN: (纸本)9789811619663;9789811619670

Unmanned aerial vehicle (UAV) base station has been proposed as a promising solution in emergency communication and supplementary communication for terrestrial networks due to its flexible layout and good mobility support. However, the dense deployment of UAV base station and ground base station brings great challenges in the configuration of neighbor cell list (NCL) during handover process. This paper presents a Cascading Bandits based Mobility Management (CBMM) algorithm for NCL configuration in the low altitude heterogeneous networks, where online learning is used to exploiting the historical handover information. In addition to the received signal strength, the cell load of each base station is also considered in the handover procedure. We aim at optimizing the configuration of NCL, so as to improve handover performance by increasing the probability of selecting the best target base station while at the same time reducing the selection delay. It is proved that the signaling overhead can be effectively reduced, since the proposed CBMM algorithm can significantly cut down the number of candidate base stations in NCL. Moreover, by ranking the candidate base stations according to their historical performance, the number of measured base stations in handover preparation phase can be effectively reduced to avoid extra delay. The simulation results of the proposed algorithm and other two existing solutions are presented to illustrate that the CBMM algorithm can achieve efficient handover management.

关键词： Handover management Heterogeneous networks Mobility management online learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Adaptive Optimization and learning for Service Systems

Adaptive Optimization and Learning for Service Systems

引用

作者： Jia, Huiwen University of Michigan

学位级别：Ph.D., Doctor of Philosophy

The primary focus of this dissertation is to develop adaptive optimization and learning models and algorithms for decision-making problems under uncertainty arising in service systems. Thanks to the accessibility and analyzability of voluminous data, the uncertainties can be better controlled by adaptively incorporating real-time information. The common theme of the problems in different chapters of this thesis embraces the structures of (i) making adaptive decisions for accomplishing a learning task and (ii) learning the nature of the uncertainty for adaptively optimizing future decisions. Specifically, we apply and innovate machine learning and reinforcement learning techniques, including support vector machines (SVM), deep neural networks (DNN), and multi-armed bandits (MAB), to solve problems that arise in modern service systems, such as transportation, resource sharing/rental services. A well-designed and reliable route planner is central to a wide range of modern transportation applications. We consider two cases, providing route recommendations to several travel requests or a single request. Providing route recommendations to a fleet of vehicles under uncertainty always results in large-scale stochastic programs, of which the solving process is time-consuming even with sophisticated decomposition algorithms. Therefore, in Chapter 2, we propose a learning-enhanced Benders decomposition (LearnBD) algorithm to reduce the solving time for two-stage stochastic programs. This algorithm can also be used for solving general two-stage stochastic programs. In Chapter 3, we design and implement an approach for providing route recommendations to one origin-destination pair via a combination of the weighted shortest path problem and deep learning with real-world transportation data. Revenue management with reusable resources finds a wide range of service systems in today's economy, such as cloud computing services, ride-hailing services, and car/bicycle rental services. T

关键词： Adaptive optimization Data-driven route planning online learning algorithms Revenue management Decision making

来源：评论

学校读者我要写书评

暂无评论

learning AIDED DYNAMIC SPECTRUM ACCESS IN DECENTRALIZED NETWORKS WITH UNKNOWN NUMBER OF SECONDARY USERS

LEARNING AIDED DYNAMIC SPECTRUM ACCESS IN DECENTRALIZED NETW...

引用

作者： SUMAN PANI Indraprastha Institute of Information Technology Delhi

学位级别：硕士

Opportunistic spectrum access (OSA) in decentralized network is a challenging problem since each unlicensed user (i. e. cognitive radio (CR)) needs to characterize frequency bands as per the occupancy statistics of multiple licensed users. In dynamic and heterogeneous networks, all CRs may not be active simultaneously and can leave or join the network at any time. This makes OSA more challenging especially in the decentralized network where active CRs do not have any knowledge of other CRs in the network. In this thesis, a new decision making policy (DMP) using online learning algorithms has been proposed for characterization of frequency bands and orthogonalization of CRs into different but optimal bands. The proposed DMP, when implemented at all CRs in decentralized network, leads to an order-optimal policy. There are two main underlying strategies 1) learning the subband availability, which helps to reduce the long waiting time due to occupancy by primary user(PU), and 2) learning the efficiency of rank, which helps in reducing collision among the secondary users(SUs). For the fixed number of SUs, the loss in throughput decreases by 63% in case of 4 SUs and 46.6% in case of 6 SUs compared to existing state-of-art DMPs, Musical Chair(MC). For varying number of SUs, the loss in throughput decreases by 81.5% in case of 4 SUs and 84.8% in case of 6 SUs compared to MC.

关键词： opportunistic spectrum access online learning algorithms decentralize network unsynchronized cognitive radios

来源：评论

学校读者我要写书评

暂无评论

Hybrid model for predicting anomalous large passenger flow in urban metros

引用

IET INTELLIGENT TRANSPORT SYSTEMS 2020年第14期14卷 1987-1996页

作者： Zheng, Zhihao Ling, Ximan Wang, Pu Xiao, Jianhe Zhang, Fan Cent South Univ Sch Traff & Transportat Engn Rail Data Res & Applicat Key Lab Hunan Prov 22 South Shaoshan Rd Changsha 410075 Hunan Peoples R China Chinese Acad Sci Shenzhen Inst Adv Technol 1068 Xueyuan Ave Shenzhen 518055 Guangdong Peoples R China

Machine learning models have been widely adopted for passenger flow prediction in urban metros;however, the authors find machine learning models may underperform under anomalous large passenger flow conditions. In this study, they develop a prediction framework that combines the advantage of complex network models in capturing the collective behaviour of passengers and the advantage of online learning algorithms in characterising rapid changes in real-time data. The proposed method considerably improves the accuracy of passenger flow prediction under anomalous conditions. This study can also serve as an exploration of interdisciplinary methods for transportation research.

关键词： learning (artificial intelligence) transportation complex networks rail traffic real-time systems behavioural sciences transportation real-time data online learning algorithms collective behaviour urban metros hybrid model anomalous conditions passenger flow prediction complex network models prediction framework anomalous large passenger flow conditions machine learning models

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：