Showing papers on "Sequential probability ratio test published in 2006"

Error control for probabilistic model checking

Abstract: We introduce a framework for expressing correctness guarantees of model-checking algorithms. The framework allows us to qualitatively compare different solution techniques for probabilistic model checking, both techniques based on statistical sampling and numerical computation of probability estimates. We provide several new insights into the relative merits of the different approaches. In addition, we present a new statistical solution method that can bound the probability of error under any circumstances by sometimes reporting undecided results. Previous statistical solution methods could only bound the probability of error outside of an “indifference region.”

Optimization of sample size in controlled experiments: the CLAST rule.

Abstract: Sequential rules are explored in the context of null hypothesis significance testing. Several studies have demonstrated that the fixed-sample stopping rule, in which the sample size used by researchers is determined in advance, is less practical and less efficient than sequential stopping rules. It is proposed that a sequential stopping rule called CLAST (composite limited adaptive sequential test) is a superior variant of COAST (composite open adaptive sequential test), a sequential rule proposed by Frick (1998). Simulation studies are conducted to test the efficiency of the proposed rule in terms of sample size and power. Two statistical tests are used: the one-tailed t test of mean differences with two matched samples, and the chi-square independence test for twofold contingency tables. The results show that the CLAST rule is more efficient than the COAST rule and reflects more realistically the practice of experimental psychology researchers.

A Sequential Vehicle Classifier for Infrared Video using Multinomial Pattern Matching

Abstract: Vehicle classification is a challenging problem, since vehicles can take on many different appearances and sizes due to their form and function, and the viewing conditions. The low resolution of uncooled-infrared video and the large variability of naturally occurring environmental conditions can make this an even more difficult problem. We develop a multilook fusion approach for improving the performance of a single look system. Our single look approach is based on extracting a signature consisting of a histogram of gradient orientations from a set of regions covering the moving object. We use the multinomial pattern matching algorithm to match the signature to a database of learned signatures. To combine the match scores of multiple signatures from a single tracked object, we use the sequential probability ratio test. Using real infrared data we show excellent classification performance, with low expected error rates, when using at least 25 looks.

A bayesian stopping rule for a single arm study : With a case study of stem cell transplantation

Abstract: Continuous monitoring of treatment failures is an important issue in clinical studies of a single experimental treatment for high risk therapy such as hematopoietic stem cell transplantation. The sequential probability ratio test (SPRT) of Wald in 1947 and various alternative stopping rules have been proposed for sequential monitoring of adverse events. It is natural to use prior information to improve stopping rules and statistical analysis. A Bayesian stopping rule is developed and applied to an example of an umbilical cord blood transplant study performed at the University of Minnesota. Two strata, based on the number of nucleated cells per kg recipient body weight (the 'dose') are monitored separately and different rules are constructed for each stratum using different prior distributions. It is believed that patients in the lower dose group have a greater chance of graft failure than those in the higher dose group. A program, written in R, is also presented for calculating the stopping rule using the prior beliefs. The program is an improvement upon existing programs and it can be used for larger studies.

The Sequential Mesh Test for A Proportion

Abstract: This paper proposed a new sampling plan, the sequential mesh test, in order to overcome the disadvantages of the widely used Sequential Probability Ratio Test (SPRT). The main idea of the new plan is to divide a SPRT test problem into a series of subproblems by inserting a number of test points. The paper presented in detail how to realize the new plan, and showed that the new plan has a much better control of the sample number than SPRT does. Finally it showed that the new method is also more powerful than IEC 1123 the well-known international standard for sampling inspection.

Adaptive Track Detection for Multi-Static Active Sonar Systems

Abstract: Multi-static active sonar systems detect contacts of interest by transmitting coherent waveforms and detecting the echoes on one or more receiving sensors. When a target of interest is in a region where its echoes are detectable by more than one receiver it can, in general, be declared sooner by combining the measurements from all sensors. The track detection schemes used in active sonar systems are often based on the Wald Sequential Probability Ratio Test (SPRT) and take as input the amplitudes of the target echoes associated to the track and where the statistical models for the amplitude of a target echo usually depend on a signal-to-noise ratio (SNR) parameter. The Multi-Static Adaptive Track Detector (MSATD) is an SPRT based track detection scheme that uses estimates of target aspect derived from track state estimates and a model of bi-static target strength to adapt the parameters in the distribution for target echo amplitude. Essentially, the SUM detector is modified to use different values for SNR parameter at each sensor. The SNR parameters are determined using a model of bistatic target strength and estimates of the target aspect observed by each sensor computed from the current track state estimate. The theoretical improvement in system track detection performance (i.e., probability of detection and latency) afforded by the proposed method is also presented; theoretically exact expressions for probability of detection and latency are evaluated numerically for all three track detection schemes for a system of one source and two receivers

Joint sequential detection and trajectory estimation with radar applications

Abstract: The problem of signal detection and trajectory estimation of a dynamic system when a variable number of measurements can be taken is here considered. A sequential probability ratio test (SPRT) when the parameter space has infinite cardinality is proposed for the detection problem while trajectory estimation relies upon a maximum-a-posteriori (MAP) estimate. The computational costs of the proposed algorithm, whose statistics are computed through a dynamic programming (DP) algorithm, are considered and applications to radar surveillance problems are inspected.

Sequential probability ratio test (SPRT) for dynamic radiation level determination – preliminary assessment

Abstract: A Sequential Probability Ratio Test (SPRT) algorithm for reliable and fast determination of a relative radiation level in a field environment has been developed. The background and the radioactive anomaly are assumed to follow the normal and Poisson distributions, respectively. The SPRT formulation has been derived and simplified based on these assumptions. The preliminary evaluation suggests that the algorithm, while offering confident estimations for the log-scaled radiation level, promises the additional advantage of reduction in sampling sizes, particularly in areas with a high radiation level.

On Sequential On-Line Outlier Detection and a Linescan Application

Abstract: Industrial quality monitoring is increasing rapidly, and challenging signal environments with requirement of steady performance pose conflicting demands to on-line tests. The sequential probability ratio test (SPRT) and the Kalman filter (KF) are proposed as two tools for detection and recognition-oriented signal processing. A modified sequential test is suggested and applied to a linescan problem.

Optimal Two-Stage Sequential Sampling Plans by Attributes

Abstract: Acceptance sampling plans have been widely used in statistical quality control for several decades. However, when nearly perfect quality is needed, their practicability is questioned by practitioners because of required large sample sizes. Moreover, the majority of well-known sampling plans allow nonconforming items in a sample, and this contradicts the generally accepted “zero defect” paradigm. Sequential sampling plans, introduced by Wald [7], assure the lowest possible sample size. Thus, they are applicable especially for sampling products of high quality. Unfortunately, their design is rather complicated. In the paper we propose a simple, and easy to design, special case of sequential sampling plans by attributes, named CSeq-1 sampling plans, having acceptance numbers not greater than one. We analyze the properties of these plans, and compare them to the properties of other widely-used sampling procedures.

Comparison of limit standards using a sequential probability ratio test

Abstract: We continue our research into the comparison, via simulation experiments, of a stochastic system to a limit standard. A limit standard is defined as a maximum and/or minimum standard. We have found that evaluation methods using proportions provide a statistically valid comparison of this family of standards. In this paper, we present the use of Wald's sequential probability ratio test (SPRT) as a comparison method. We define a fully sequential analysis procedure, as well as initial test results.

Leak Detection in a Pipeline Using Modified Line Volume Balance and Sequential Probability Tests

Abstract: The use of statistical tools to improve the decision aspect of leak detection is becoming a common practice in the area of computer pipeline monitoring. Among these tools, the sequential probability ratio test is one of the most named techniques used by commercial leak detection systems [1]. This decision mechanism is based on the comparison of the estimated probabilities of leak or no leak observed from the pipeline data. This paper proposes a leak detection system that uses a simplified statistical model for the pipeline operation, allowing a simple implementation in the pipeline control system [2]. Applying linear regression to volume balance and average pipeline pressure signals, a statistically corrected volume balance signal with reduced variance is introduced. Its expected value is zero during normal operation whereas it equals the leak flow under a leak condition. Based on the corrected volume balance, differently configured sequential probability ratio tests (SPRT) to extend the dynamic range of detectable leak flow are presented. Simplified mathematical expressions are obtained for several system performance indices, such as spilled volume until detection, time to leak detection, minimum leak flow detected, etc. Theoretical results are compared with leak simulations on a real oil pipeline. A description of the system tested over a 500 km oil pipeline is included, showing some real data results.Copyright © 2006 by ASME

Parametrical Recognition Algorithm Synthesis Based on the Sequential Analysis

Abstract: In this paper the results of the theory of sequential analysis applying for some recognition tasks are given