David E. Booth

A Neural Network Approach to the Detection of Nuclear Material Losses

A series of repeated nuclear material balances forms a time series of often autocorrelated observations. Outliers, deviations from an in-control production process or time series pattern, indicate an out-of-control situation relative to the process norm. In this paper various methods, especially neural networks, will be examined with respect to their use to detect nuclear material diversions or losses more rapidly and accurately than currently used methods. The neural network technique will be enhanced with the use of a simulation computer program for creating the training data set. This simulation approach provides the opportunity of including outliers of various types in a data set for training the neural network because an actual process data set used for training possibly may not have outliers. In this paper, the methods will be compared on their ability to identify outliers and reduce false alarms. These methods were tested on data sets of nuclear material balances with known removals, and the result...

On robust partial discriminant analysis as a decision-making tool with clinical and analytical chemical data

Classification is one of the fundamental goals of science and is basic to the diagnosis of disease. Unfortunately, classifying objects (e.g., patients) on the basis of clinical and/or laboratory experimental observations into various groups can be difficult when the groups overlap or contain outlying points. Recently, Broffitt, Randles, and co-workers proposed a procedure, robust partial discriminant analysis (RPDA) for dealing with such problems, but testing of the procedure was limited to Monte Carlo simulation. In this study, RPDA was applied to real data, in order to compare its effectiveness with ordinary discriminant analysis, as well as to determine if RPDA was a suitable procedure to use to classify chemical compounds on the basis of experimental observations and as a tool in the diagnosis of disease (in particular, multiple sclerosis and thyrotoxicosis), with data based on experimental and clinical observations. The resulting RPDA classifications were an improvement over those obtained from ordinary discriminant analysis.

Robust regression-based analysis of drug–nucleic acid binding

Outlier detection can be very important in analyzing data from Scatchard plots. In this study, a robust (outlier-resistant) regression procedure was used in conjunction with a Scatchard plot to study the binding of the methylphenazinium cation with double-stranded DNA. The procedures, their results, and their advantages are discussed.

A ROBUST SMOOTHING APPROACH TO STATISTICAL PROCESS CONTROL

It has previously been shown that smoothing algorithms can provide the basis for methods to detect nuclear material losses and moreover can also provide a general approach to industrial statistical process control. The present paper extends this result by showing that a set of robust smoothers also produces methods that can be used in statistical process control. Further, it is shown that these smoothers are somewhat more sensitive to out of control points than those methods previously studied. The methods are successfully illustrated on chemical process data.

The Use of Robust Smoothers in Nuclear Material Safeguards

It has previously been shown that smoothing algorithms can provide the basis for a method to detect nuclear material diversions and losses and moreover can also provide a general approach to industrial statistical process control. The present paper extends this result by showing that a set of robust smoothers also produces equivalent methods that can be used in nuclear material safeguards algorithms. Further, it is shown that these smoothers are somewhat more sensitive to loss points than the previously studied smoothers. The method is illustrated on real data.

A Prototype System Developed for Digital Rights Management in Electronic Commerce

ABSTRACT The Internet provides a new way of doing business, but ease of copying and of sharing valuable digital information illegally across the Internet undermines many viable business models. This paper investigates Digital Rights Management (DRM) as a means to provide safe protection and proper delivery of digital contents through the Information highway. First, we briefly summarize the current endeavor of DRM technical development and its standardization process in key technical working groups. Then, the paper provides a generic architecture for a DRM framework and shows the implementation of a prototype DRM system incorporating key conceptual and technical standardization development. This study emphasizes the importance of developing the DRM architecture that provides the proper protection and safe transformation of digital contents in electronic commerce.

Using Polynomial Smoothing and Data Bounding for the Detection of Adverse Process Changes in a Chemical Process

published in Advance ACS Abstracts, April 1, 1994. 0095-2338/94/1634-0881

A Robust Multivariate Procedure for the Identification of Problem Savings and Loan Institutions

04.50/0 The goal is to discover outliers indicative of an adverse process change as early as possible while minimizing the frequency of false alarms. For example, in the case of a drifting process mean due to a gradually deteriorating part in the process, outliers would be identified while the output is still within conventional control chart limits, well before any substandard product is produced. Then the operator may make adjustments in a timely manner. Of course, a sudden malfunction in the process, such as the fracture of a part, may immediately yield poor quality product. Objectives of this Research. With this introduction, the reader may better appreciate the objectives of this research: (1) to develop outlier detection methods that are relatively simple in concept, flexible, and adaptable to the process environment; (2) to develop techniques that can be used to modify the existing control chart methodology, leading to an earlier detection of outliers indicative of an adverse process change, while reducing the probability of a false alarm by using a more realistic process model. B. APPLICATIONS TO INDUSTRIAL PROCESS CONTROL AND SOME PRIOR RESEARCH used the generalized M estimator (GM) procedure of Denby and Marting to fit a time series model to a set of observations from a stable production process. The resulting model is then used as the basis for detecting changes in the process. The GM method models the process more accurately and gives information on the type of change, which is helpful in finding the cause of the problem, by identifying additive and innovative outliers. This is extremely useful information in determining how to return the production process to the incontrol state. However, the GM procedure is limited to those processes that can be modeled as a pth-order autoregressive [AR(p)] time series. Booth and Isenhour6 successfully used the GM method for the early detection of poisoning of a platinum catalyst used for the oxidation of ammonia to nitric acid. BoothS also used the method for the early discovery of nuclear material losses. Prasad et al.*O recently tested Chen and Liu’s8 joint estimation procedure, an extension of the GM idea with Q 1994 American Chemical Society 882 J. Chem. In5 Comput. Sci., Vol. 34, No. 4, 1994 SEBASTIAN ET AL. satisfactory results on chemical process data sets with known outliers. It is more discriminating in identifying outlier types and can deal with any type of time series model, not only