R. Kurt Ungar

Radiostrontium and radium analysis in low-level environmental samples following a multi-stage semi-automated chromatographic sequential separation

Strontium isotopes, (89)Sr and (90)Sr, and (226)Ra being radiotoxic when ingested, are routinely monitored in milk and drinking water samples collected from different regions in Canada. In order to monitor environmental levels of activity, a novel semi-automated sensitive method has been developed at the Radiation Protection Bureau of Health Canada (Ottawa, Canada). This method allows the separation and quantification of both (89)Sr and (90)Sr and has also been adapted to quantify (226)Ra during the same sample preparation procedure. The method uses a 2-stage purification process during which matrix constituents, such as magnesium and calcium that are rich in milk, are removed as well as the main beta-interferences (e.g., (40)K, (87)Rb, (134)Cs, (137)Cs, and (140)Ba). The first purification step uses strong cation exchange (SCX) chromatography with commercially available resins. In a second step, fractions containing the radiostrontium analytes are further purified using high-performance ion chromatography (HPIC). While (89)Sr is quantified by Cerenkov counting immediately after the second purification stage, the same vial is counted again after a latent period of 10-14 days to quantify the (90)Sr activity based on (90)Y ingrowth. Similarly, the activity of (226)Ra, which is separated by SCX only, is determined via the emanation of (222)Rn in a 2-phase aqueous/cocktail system using liquid scintillation counting. The minimum detectable concentration (MDC) for (89)Sr and (90)Sr for a 200 min count time at 95% confidence interval is 0.03 and 0.02 Bq/L, respectively. The MDC for (226)Ra for a 100 min count time is 0.002 Bq/L. Semi-annual intercomparison samples from the USA Department of Energy Mixed Analyte Performance Evaluation Program (MAPEP) were used to validate the method for (89)Sr and (90)Sr. Spiked water samples prepared in-house and from International Atomic Energy Agency (IAEA) were used to validate the (226)Ra assay.

Full border identification for reduction of training sets

Border identification (BI) was previously proposed to help learning systems focus on the most relevant portion of the training set so as to improve learning accuracy. This paper argues that the traditional BI implementation suffers from a serious limitation: it is only able to identify partial borders. This paper proposes a new BI method called Progressive Border Sampling (PBS), which addresses this limitation by borrowing ideas from recent research on Progressive Sampling. PBS progressively learns optimal borders from the entire training sets by, first, identifying a full border, thus, avoiding the limitation of the traditional BI method, and, second, by incrementing the size of that border until it converges to an optimal sample, which is smaller than the original training set. Since PBS identifies the full border, it is expected to discover more optimal samples than traditional BI. Our experimental results on the selected 30 benchmark datasets from the UCI repository show that, indeed, in the context of classification, PBS is more successful than traditional BI at reducing the size of the training sets and optimizing the accuracy results.

Anomaly detection in gamma ray spectra: A machine learning perspective

With Canadian security and the safety of the general public in mind, physicists at Health Canada (HC) have begun to develop techniques to identify persons concealing radioactive material that may represent a threat to attendees at public gatherings, such as political proceedings and sporting events. To this end, Health Canada has initiated field trials that include the deployment of gamma-ray spectrometers. In particular, a series of these detectors, which take measurements every minute and produce 1,024 channel gamma-ray spectrum, were deployed during the Vancouver 2010 olympics. Simple computerized statistics and human expertise were used as the primary line of defence. More specifically, if a measured spectrum deviated significantly from the background, an internal alarm was sounded and an HC physicist undertook further analysis into the nature of the alarming spectrum. This strategy, however, lead to a significant number of costly and time consuming false positives. This research applies sophisticated machine learning algorithms to reduce the number of false positives to an acceptable level, the results of which are detailed in this paper. In addition, we emphasize the primary findings of our work and highlight avenues available to further improve upon our current results.

Machine learning for radioxenon event classification for the Comprehensive Nuclear-Test-Ban Treaty.

A method of weapon detection for the Comprehensive nuclear-Test-Ban-Treaty (CTBT) consists of monitoring the amount of radioxenon in the atmosphere by measuring and sampling the activity concentration of (131m)Xe, (133)Xe, (133m)Xe, and (135)Xe by radionuclide monitoring. Several explosion samples were simulated based on real data since the measured data of this type is quite rare. These data sets consisted of different circumstances of a nuclear explosion, and are used as training data sets to establish an effective classification model employing state-of-the-art technologies in machine learning. A study was conducted involving classic induction algorithms in machine learning including Naïve Bayes, Neural Networks, Decision Trees, k-Nearest Neighbors, and Support Vector Machines, that revealed that they can successfully be used in this practical application. In particular, our studies show that many induction algorithms in machine learning outperform a simple linear discriminator when a signal is found in a high radioxenon background environment.

Nuclear event zero-time calculation and uncertainty evaluation.

It is important to know the initial time, or zero-time, of a nuclear event such as a nuclear weapons test, a nuclear power plant accident or a nuclear terrorist attack (e.g. with an improvised nuclear device, IND). Together with relevant meteorological information, the calculated zero-time is used to help locate the origin of a nuclear event. The zero-time of a nuclear event can be derived from measured activity ratios of two nuclides. The calculated zero-time of a nuclear event would not be complete without an appropriately evaluated uncertainty term. In this paper, analytical equations for zero-time and the associated uncertainty calculations are derived using a measured activity ratio of two nuclides. Application of the derived equations is illustrated in a realistic example using data from the last Chinese thermonuclear test in 1980.

Instance Selection by Border Sampling in Multi-class Domains

Instance selection is a pre-processing technique for machine learning and data mining. The main problem is that previous approaches still suffer from the difficulty to produce effective samples for training classifiers. In recent research, a new sampling technique, called Progressive Border Sampling (PBS), has been proposed to produce a small sample from the original labelled training set by identifying and augmenting border points. However, border sampling on multi-class domains is not a trivial issue. Training sets contain much redundancy and noise in practical applications. In this work, we discuss several issues related to PBS and show that PBS can be used to produce effective samples by removing redundancies and noise from training sets for training classifiers. We compare this new technique with previous instance selection techniques for learning classifiers, especially, for learning Naive Bayes-like classifiers, on multi-class domains except for one binary case which was for a practical application.

Monte Carlo simulations of NaI(Tl) spectra for measurements of semi-infinite plumes.

For the past 10 y Health Canada has operated a Fixed Point Surveillance Network of NaI(Tl) detectors across Canada. Deployed for both emergency response and daily monitoring of airborne radiation in the environment, a spectral stripping method allowed measurement of certain isotopes well below the ambient dose rate. These include (133)Xe, (135)Xe and (41)Ar, typical of emissions from operating nuclear reactors. In an effort to increase the number of isotopes measured at these low levels a new technique of spectral fitting using spectral templates is being implemented. However, this requires very accurate spectral templates that can be difficult or impossible to obtain empirically for environmental measurements of airborne radio-isotopes. Therefore, a method of efficiently using Monte Carlo techniques to create these templates was developed.

Cascading customized naïve bayes couple

Naive Bayes (NB) is an efficient and effective classifier in many cases However, NB might suffer from poor performance when its conditional independence assumption is violated While most recent research focuses on improving NB by alleviating the conditional independence assumption, we propose a new Meta learning technique to scale up NB by assuming an altered strategy to the traditional Cascade Learning (CL) The new Meta learning technique is more effective than the traditional CL and other Meta learning techniques such as Bagging and Boosting techniques while maintaining the efficiency of Naive Bayes learning.