Statistical learning theory and randomized algorithms for control

The topic of the present article is the use of randomized algorithms to solve some problems in control system designs that are perceived to be "difficult". A brief introduction is given to the notions of computational complexity that are pertinent to the present discussion, and then some problems in control system analysis and synthesis that are difficult in a complexity-theoretic sense are described. Some of the elements of statistical learning theory, which forms the basis of the randomized approach, are briefly described. Finally, these two sets of ideas are brought together to show that it is possible to construct efficient randomized algorithms for each of the difficult problems discussed by using the ideas of statistical learning theory. A real-life design example of synthesizing a first-order controller for the longitudinal stabilization of an unstable fighter aircraft is then presented to show that the randomized approach can be quite successful in tackling a practical problem.

[1]  R. Stengel,et al.  Technical notes and correspondence: Stochastic robustness of linear time-invariant control systems , 1991 .

[2]  M. Morari,et al.  Computational complexity of μ calculation , 1994, IEEE Trans. Autom. Control..

[3]  V. Vapnik,et al.  Necessary and Sufficient Conditions for the Uniform Convergence of Means to their Expectations , 1982 .

[4]  Marek Karpinski,et al.  Polynomial bounds for VC dimension of sigmoidal neural networks , 1995, STOC '95.

[5]  Marek Karpinski,et al.  Polynomial Bounds for VC Dimension of Sigmoidal and General Pfaffian Neural Networks , 1997, J. Comput. Syst. Sci..

[6]  John C. Doyle,et al.  Control problems and the polynomial time hierarchy , 1995, Proceedings of 1995 34th IEEE Conference on Decision and Control.

[7]  J. Steele.  Empirical Discrepancies and Subadditive Processes , 1978 .

[8]  David Haussler,et al.  Decision Theoretic Generalizations of the PAC Model for Neural Net and Other Learning Applications , 1992, Inf. Comput..

[9]  Robert F. Stengel,et al.  The use of random search and genetic algorithms to optimize stochastic robustness functions , 1994, Proceedings of 1994 American Control Conference - ACC '94.

[10]  W. Hoeffding.  Probability Inequalities for sums of Bounded Random Variables , 1963 .

[11]  M. Safonov.  Stability margins of diagonally perturbed multivariable feedback systems , 1982 .

[12]  Stephen A. Cook,et al.  The complexity of theorem-proving procedures , 1971, STOC.

[13]  Christopher L. DeMarco,et al.  The computational complexity of approximating the minimal perturbation scaling to achieve instability in an interval matrix , 1994, Math. Control. Signals Syst..

[14]  B. Anderson,et al.  Output feedback stabilization and related problems-solution via decision methods , 1975 .

[15]  Svatopluk Poljak,et al.  Checking robust nonsingularity is NP-hard , 1993, Math. Control. Signals Syst..

[16]  A. Nemirovskii.  Several NP-hard problems arising in robust stability analysis , 1993 .

[17]  R. Tempo,et al.  Probabilistic robustness analysis: explicit bounds for the minimum number of samples , 1997 .

[18]  Pramod P. Khargonekar,et al.  Randomized algorithms for robust control analysis and synthesis have polynomial complexity , 1996, Proceedings of 35th IEEE Conference on Decision and Control.

[19]  Thomas Kailath,et al.  Linear Systems , 1980 .

[20]  John C. Doyle.  Analysis of Feedback Systems with Structured Uncertainty , 1982 .

[21]  H. Chernoff.  A Measure of Asymptotic Efficiency for Tests of a Hypothesis Based on the sum of Observations , 1952 .

[22]  J. Tsitsiklis,et al.  NP-Hardness of Some Linear Control Design Problems , 1997 .

[23]  Eduardo D. Sontag,et al.  Finiteness results for sigmoidal “neural” networks , 1993, STOC.