基于SMO的层次型1-FSVM算法

针对序贯最小优化(SMO)训练算法具有计算速度快、无内负荷的特点,将其移植到模糊一类支持向量机(1-FSVM)中。1-FSVM算法融入层次型偏二叉树结构进行逐步聚类以加快训练速度,并对每个输入向量赋予不同权值以达到准确的分类效果。应用于光识别手写数字集和车牌定位的结果表明,1-FSVM算法具有较高的检测率与较快的检测速度。     

An adaptive peer-to-peer live streaming system with incentives for resilience

Recently, there have been a lot of research efforts on peer-to-peer (P2P) live streaming services. P2P systems can be easily deployed since a participating peer’s resources (i.e., upload link bandwidth) can be exploited to distribute contents. However, how to adapt to leaving peers and how to encourage peers to contribute resources voluntarily are still challenging issues. In this paper, we propose Climber, an adaptive P2P live streaming system with incentives for resilience. Climber is based on the hybrid structure of a tree and a mesh, so as to achieve self-improvement and adaptation to users’ dynamic joining and leaving. Moreover, Climber substantiates an incentive mechanism that provides better resilience for peers with more upload bandwidth allocated. Simulation results reveal that Climber significantly reduces the topology maintenance cost compared to SplitStream and NICE-PRM. Also, simulation and analytical results verify that Climber can bound the level of disruption by dynamically adapting to the user churning rate.     

Multi-goal Q-learning of cooperative teams

This paper studies a multi-goal Q-learning algorithm of cooperative teams. Member of the cooperative teams is simulated by an agent. In the virtual cooperative team, agents adapt its knowledge according to cooperative principles. The multi-goal Q-learning algorithm is approached to the multiple learning goals. In the virtual team, agents learn what knowledge to adopt and how much to learn (choosing learning radius). The learning radius is interpreted in Section 3.1. Five basic experiments are manipulated proving the validity of the multi-goal Q-learning algorithm. It is found that the learning algorithm causes agents to converge to optimal actions, based on agents’ continually updated cognitive maps of how actions influence learning goals. It is also proved that the learning algorithm is beneficial to the multiple goals. Furthermore, the paper analyzes how sensitive the learning performance is affected by the parameter values of the learning algorithm.     

An improved adaptive genetic algorithm based on hormone modulation mechanism for job-shop scheduling problem

An improved adaptive genetic algorithm (IAGA) for solving the minimum makespan problem of job-shop scheduling problem (JSP) is presented. Though the traditional genetic algorithm (GA) exhibits implicit parallelism and can retain useful redundant information about what is learned from previous searches by its representation in individuals in the population, yet GA may lose solutions and substructures due to the disruptive effects of genetic operators and is not easy to regulate GA’s convergence. The proposed IAGA is inspired from hormone modulation mechanism, and then the adaptive crossover probability and adaptive mutation probability are designed. The proposed IAGA is characterized by simplifying operations, high search precision, overcoming premature phenomenon and slow evolution. The proposed method by employing operation-based encoding is effectively applied to solve a dynamic job-shop scheduling problem (DJSP) and a complicated contrastive experiment of JSP in manufacturing system. Meanwhile, in order to ensure to create a feasible solution, a new method for crossover operation is adopted, named, partheno-genetic operation (PGO). The computational results validate the effectiveness of the proposed IAGA, which can not only find optimal or close-to-optimal solutions but can also obtain both better and more robust results than the existing genetic algorithms reported recently in the literature. By employing IAGA, machines can be used more efficiently, which means that tasks can be allocated appropriately, production efficiency can be improved, and the production cycle can be shortened efficiently.     

Adaptive neural control for uncertain stochastic nonlinear strict-feedback systems with time-varying delays: A Razumikhin functional method

This paper addresses the problem of adaptive neural control for a class of uncertain stochastic nonlinear strict-feedback systems with time-varying delays. A novel adaptive neural control scheme is presented for this class of systems, based on a combination of the Razumikhin functional approach, the backstepping technique and the neural network (NN) parameterization. The proposed adaptive controller guarantee that all the error variables are 4-Moment semi-globally uniformly ultimately bounded in a compact set while the system output converges to a small neighborhood of the reference signal. Two simulation examples are given to demonstrate the effectiveness of the proposed control schemes.     

Deterministic convergence of conjugate gradient method for feedforward neural networks

Conjugate gradient methods have many advantages in real numerical experiments, such as fast convergence and low memory requirements. This paper considers a class of conjugate gradient learning methods for backpropagation neural networks with three layers. We propose a new learning algorithm for almost cyclic learning of neural networks based on PRP conjugate gradient method. We then establish the deterministic convergence properties for three different learning modes, i.e., batch mode, cyclic and almost cyclic learning. The two deterministic convergence properties are weak and strong convergence that indicate that the gradient of the error function goes to zero and the weight sequence goes to a fixed point, respectively. It is shown that the deterministic convergence results are based on different learning modes and dependent on different selection strategies of learning rate. Illustrative numerical examples are given to support the theoretical analysis.     

Sliding mode control for polytopic differential inclusion systems

The stabilization problem of polytopic differential inclusion (PDI) systems is investigated by using sliding mode control. Sliding surface is designed and sufficient conditions for asymptotic stability of sliding mode dynamics are derived. A novel feedback law is established to make the state of system reach the sliding surface in a finite time. Finally, an example is given to illustrate the validity of the proposed design.     

Sampling period scheduling of networked control systems with multiple-control loops

This paper addresses the sampling period scheduling of Networked Control Systems (NCSs) with multiple control loops. The generalized exponential function is employed to describe Integral Absolute Error (IAE) performance versus sampling period by Truetime toolbox under Matlab environment, and the sampling periods are scheduled to obtain the optimal integrated performance based on Kuhn–Tucker Theorem, which are subject to the stability of every control loop and the bandwidth on available network resource. Numerical examples are given to show the effectiveness of our method.