Distribution of the Length of the Longest Significance Run on a Bernoulli Net and Its Applications

We consider the length of the longest significance run in a (two-dimensional) Bernoulli net and derive its asymptotic limit distribution. Our results can be considered as generalizations of known theorems in significance runs. We give three types of theoretical results: (1) reliability-style lower and upper bounds, (2) Erdös–Rényi law, and (3) the asymptotic limit distribution. To understand the rate of convergence to the asymptotic distributions, we carry out numerical simulations. The convergence rates in a variety of situations are presented. To understand the relation between the length of the longest significance run(s) and the success probability p, we propose a dynamic programming algorithm to implement simultaneous simulations. Insights from numerical studies are important for choosing the values of design parameters in a particular application, which motivates this article. The distribution of the length of the longest significance run in a Bernoulli net is critical in applying a multiscale methodology in image detection and computational vision. Approximation strategies to some critical quantities are discussed.

[1]  E. J. Burr,et al.  Longest run of consecutive observations having a specified attribute , 1961 .

[2]  D. Walkup,et al.  Association of Random Variables, with Applications , 1967 .

[3]  A. Rényi,et al.  On a new law of large numbers , 1970 .

[4]  L. V. Osipov.  On Probabilities of Large Deviations for Sums of Independent Random Variables , 1973 .

[5]  Shimon Ullman,et al.  Structural Saliency: The Detection Of Globally Salient Structures using A Locally Connected Network , 1988, [1988 Proceedings] Second International Conference on Computer Vision.

[6]  M. Waterman,et al.  The Erdos-Renyi Law in Distribution, for Coin Tossing and Sequence Matching , 1990 .

[7]  Markos V. Koutras,et al.  Bounds for reliability of consecutive k-within-m-out-of-n:F systems , 1993 .

[8]  Joseph Naus,et al.  Power of the scan statistic in detecting a changed segment in a Bernoulli sequence , 1994 .

[9]  J. Ben Atkinson,et al.  Modeling and Analysis of Stochastic Systems , 1996 .

[10]  Marco Muselli,et al.  On convergence properties of pocket algorithm , 1997, IEEE Trans. Neural Networks.

[11]  Marco Muselli,et al.  Useful inequalities for the longest run distribution , 2000 .

[12]  Małgorzata Roos,et al.  Runs and Scans With Applications , 2001 .

[13]  Andrew D. Barbour,et al.  Compound Poisson approximation: a user's guide , 2001 .

[14]  Michael R. Chernick,et al.  Runs and Scans With Applications , 2002, Technometrics.

[15]  Tze Leung Lai,et al.  Saddlepoint approximations and nonlinear boundary crossing probabilities of Markov random walks , 2002 .

[16]  Xiaoming Huo,et al.  Multiscale significance run: realizing the 'most powerful' detection in noisy images , 2003, The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003.

[17]  F. JAMESC.,et al.  ON EXACT AND LARGE DEVIATION APPROXIMATION FOR THE DISTRIBUTION OF THE LONGEST RUN IN A SEQUENCE OF TWO-STATE MARKOV DEPENDENT TRIALS , 2003 .

[18]  D. Donoho,et al.  Adaptive multiscale detection of filamentary structures embedded in a background of uniform random points , 2003 .

[19]  Lionel Moisan,et al.  Meaningful Alignments , 2000, International Journal of Computer Vision.

[20]  P. Révész,et al.  On the length of the longest head run , 2004 .