Extreme learning machine for regression and multiclass classification

Due to the simplicity of their implementations, least square support vector machine (LS-SVM) and proximal support vector machine (PSVM) have been widely used in binary classification applications. The conventional LS-SVM and PSVM cannot be used in regression and multiclass classification applications directly, although variants of LS-SVM and PSVM have been proposed to handle such cases. This paper shows that both LS-SVM and PSVM can be simplified further and a unified learning framework of LS-SVM, PSVM, and other regularization algorithms referred to extreme learning machine (ELM) can be built. ELM works for the "generalized" single-hidden-layer feedforward networks (SLFNs), but the hidden layer (or called feature mapping) in ELM need not be tuned. Such SLFNs include but are not limited to SVM, polynomial network, and the conventional feedforward neural networks. This paper shows the following: 1) ELM provides a unified learning platform with a widespread type of feature mappings and can be applied in regression and multiclass classification applications directly; 2) from the optimization method point of view, ELM has milder optimization constraints compared to LS-SVM and PSVM; 3) in theory, compared to ELM, LS-SVM and PSVM achieve suboptimal solutions and require higher computational complexity; and 4) in theory, ELM can approximate any target continuous function and classify any disjoint regions. As verified by the simulation results, ELM tends to have better scalability and achieve similar (for regression and binary class cases) or much better (for multiclass cases) generalization performance at much faster learning speed (up to thousands times) than traditional SVM and LS-SVM.     

Hybrid function projective synchronization of chaotic systems with uncertain time-varying parameters via Fourier series expansion

In this paper, the hybrid function projective synchronization (HFPS) of different chaotic systems with uncertain periodically time-varying parameters is carried out by Fourier series expansion and adaptive bounding technique. Fourier series expansion is used to deal with uncertain periodically time-varying parameters. Adaptive bounding technique is used to compensate the bound of truncation errors. Using the Lyapunov stability theory, an adaptive control law and six parameter updating laws are constructed to make the states of two different chaotic systems asymptotically synchronized. The control strategy does not need to know the parameters thoroughly if the time-varying parameters are periodical functions. Finally, in order to verify the effectiveness of the proposed scheme, the HFPS between Lorenz system and Chen system is completed successfully by using this scheme.     

Globally asymptotic stabilization for nonlinear time-delay systems with ISS inverse dynamics

The output feedback stabilization is considered for a class of nonlinear time-delay systems with inverse dynamics in this paper.An appropriate state observer is constructed for the unmeasurable system states in order to realize the control objective.By adopting the backstepping and Lyapunov-Krasovskii functional methods,a systematic design procedure for a memoryless output feedback control law is presented.It is shown that the designed controller can make the closed-loop system globally asymptotically stable while keeping all signals bounded.An illustrative example is discussed to show the effectiveness of the proposed control strategy.     

Delay‐dependent stability for Markovian genetic regulatory networks with time‐varying delays

This paper is concerned with the problem of stochastic stability analysis for a class of genetic regulatory networks with Markovian jump parameters and time‐varying delays. A delay‐dependent stability criterion is derived by using a novel mode‐dependent Lyapunov functional. The derived stability criterion is expressed in terms of linear matrix inequalities and is less conservative than the existing ones in the literature. A numerical example is provided to demonstrate the effectiveness of the proposed stability criterion. Copyright © 2011 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society     

Interactive genetic algorithms with large population and semi-supervised learning

Interactive genetic algorithms are effective methods of solving optimization problems with implicit (qualitative) criteria by incorporating a user's intelligent evaluation into traditional evolution mechanisms. The heavy evaluation burden of the user, however, is crucial and limits their applications in complex optimization problems. We focus on reducing the evaluation burden by presenting a semi-supervised learning assisted interactive genetic algorithm with large population. In this algorithm, a population with many individuals is adopted to efficiently explore the search space. A surrogate model built with an improved semi-supervised learning method is employed to evaluate a part of individuals instead of the user to alleviate his/her burden in evaluation. Incorporated with the principles of the improved semi-supervised learning, the opportunities of applying and updating the surrogate model are determined by its confidence degree in estimation, and the informative individuals reevaluated by the user are selected according to the concept of learning from mistakes. We quantitatively analyze the performance of the proposed algorithm and apply it to the design of sunglasses lenses, a representative optimization problem with one qualitative criterion. The empirical results demonstrate the strength of our algorithm in searching for satisfactory solutions and easing the evaluation burden of the user.     

An efficient initialization approach of Q-learning for mobile robots

This article demonstrates that Q-learning can be accelerated by appropriately specifying initial Q-values using dynamic wave expansion neural network. In our method, the neural network has the same topography as robot work space. Each neuron corresponds to a certain discrete state. Every neuron of the network will reach an equilibrium state according to the initial environment information. The activity of the special neuron denotes the maximum cumulative reward by following the optimal policy from the corresponding state when the network is stable. Then the initial Q-values are defined as the immediate reward plus the maximum cumulative reward by following the optimal policy beginning at the succeeding state. In this way, we create a mapping between the known environment information and the initial values of Q-table based on neural network. The prior knowledge can be incorporated into the learning system, and give robots a better learning foundation. Results of experiments in a grid world problem show that neural network-based Q-learning enables a robot to acquire an optimal policy with better learning performance compared to conventional Q-learning and potential field-based Qlearning.     

Quantum computing via the Bethe ansatz

We recognize quantum circuit model of computation as factorisable scattering model and propose that a quantum computer is associated with a quantum many-body system solved by the Bethe ansatz. As an typical example to support our perspectives on quantum computation, we study quantum computing in one-dimensional nonrelativistic system with delta-function interaction, where the two-body scattering matrix satisfies the factorisation equation (the quantum Yang–Baxter equation) and acts as a parametric two-body quantum gate. We conclude by comparing quantum computing via the factorisable scattering with topological quantum computing.