Abstract
Consideration was given to the control of processing large amounts of data, provided that there are two alternative methods of processing with different and a priori unknown efficiencies. It was required to determine the most efficient method and maintain its preferable use. With the use of parallel processing this may be carried out in a relatively small number of steps and actually without losses in the control performance, that is, without increasing the minimax risk. An invariant equation with a solution containing a singularity at t = 0 was previously obtained to describe the control. This solution was represented as a product with one cofactor being the density of the normal distribution which is singular at t = 0 and the other, the nonsingular one, the solution to a new equation. Numerical experiments demonstrated that this new equation offers greater possibilities for calculations. In particular, it enabled one to improve the asymptotic estimates of the minimax risk.
Similar content being viewed by others
References
Kolnogorov, A.V., Parallel Design of Robust Control in the Stochastic Environment (the Two-armed Bandit Problem), Autom. Remote Control, 2012, vol. 73, no. 4, pp. 689–701.
Vogel, W., An Asymptotic Minimax Theorem for the Two-armed Bandit Problem, Ann. Math. Stat., 1960, vol. 31, pp. 444–451.
Nazin, A.V. and Poznyak, A.S., Adaptivnyi vybor variantov (Adaptive Selection of Variants), Moscow: Nauka, 1986.
Poznyak, A.S. and Najim, K., Learning Automata and Stochastic Optimization, in Lect. Notes Control Inf. Sci., vol. 225, Berlin: Springer, 1997.
Juditsky, A., Nazin, A.V., Tsybakov, A.B., et al., Gap-free Bounds for Stochastic Multi-Armed Bandit, in Proc. 17th World Congr. Int. Federation Autom. Control, Seoul, 2008, July 6–11, pp. 11560–11563.
Witmer, J.A., Bayesian Multistage Decision Problems, Ann. Statist., 1986, vol. 14, pp. 283–297.
Lai, T.L., Levin, B., Robbins, H., et al., Sequential Medical Trials (Stopping Rules/Asymptotic Optimality), in Proc. Natl. Acad. Sci. USA, 1980, vol. 77, no. 6, pp. 3135–3138.
Kolnogorov, A.V., Problem of Two-armed Bandit for Systems with Parallel Data Processing, Probl. Peredachi Inf., 2012, vol. 48, no. 1, pp. 83–95.
Author information
Authors and Affiliations
Corresponding author
Additional information
Original Russian Text © A.V. Kolnogorov, 2014, published in Avtomatika i Telemekhanika, 2014, No. 12, pp. 42–55.
Rights and permissions
About this article
Cite this article
Kolnogorov, A.V. Robust parallel control in a random environment and data processing optimization. Autom Remote Control 75, 2124–2134 (2014). https://doi.org/10.1134/S0005117914120042
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S0005117914120042