Showing papers by "Jinchun Hu published in 2007"

Stochastic Gradient Algorithm Under (h,φ)-Entropy Criterion

Abstract: Motivated by the work of Erdogmus and Principe, we use the error (h,φ)-entropy as the supervised adaptation criterion. Several properties of the (h,φ)-entropy criterion and the connections with traditional error criteria are investigated. By a kernel estimate approach, we obtain the nonparametric estimator of the instantaneous (h,φ)-entropy. Then, we develop the general stochastic information gradient algorithm, and derive the approximate upper bound for the step size in the adaptive linear neuron training. Moreover, the (h,φ) pair are optimized to improve the performance of the proposed algorithm. For the finite impulse response identification with white Gaussian input and noise, the exact optimum φ function is derived. Finally, simulation experiments verify the results and demonstrate the noticeable performance improvement that may be achieved by the optimum (h,φ)-entropy criterion.

Measure observability by the generalized informational correlation

Abstract: Informational correlation coefficient (ICC) can be used to measure the degree of observability for a system. In this paper, we define the generalized informational correlation coefficient (GICC), which is suitable for both discrete and continuous random variables. For the case in which the probability density functions (PDFs) are regularly supersummable, we obtain the exact value of GICC. Moreover, for the linear, stochastically autonomous system, we derive the explicit formula for the degree of observability, and prove the equivalence between the proposed measure and the traditional rank condition. Finally, a simple example is given to compare the discrete state case and the continuous state case.

Satisfactory control of discrete-time linear periodic systems

Abstract: In this paper satisfactory control for discrete-time linear periodic systems is studied. Based on a suitable time-invariant state sampled reformulation, periodic state feedback controller has been designed such that desired requirements of steady state covariance, H-infinity rejection bound and regional pole assignment for the periodic system are met simultaneously. By using satisfactory control theory, the problem of satisfactory periodic controller can be transformed into a linear programming problem subject to a set of linear matrix inequalities (LMIs), and a feasible designing approach is presented via LMI technique. Numeric example validates the obtained conclusion.