Bayesian aggregation of evidence for detection and characterization of patterns in multiple noisy observations

Abstract

Effective use of Machine Learning to support extracting maximal information from limited sensor data is one of the important research challenges in robotic sensing. This thesis develops techniques for detecting and characterizing patterns in noisy sensor data. Our Bayesian Aggregation (BA) algorithmic framework can leverage data fusion from multiple low Signal-To-Noise Ratio (SNR) sensor observations to boost the capability to detect and characterize the properties of a signal generating source or process of interest.We illustrate our research with application to the nuclear threat detection domain. Developed algorithms are applied to the problem of processing the large amounts of gamma ray spectroscopy data that can be produced in real-time by mobile radiation sensors. The thesis experimentally shows BA's capability to boost sensor performance in detecting radiation sources of interest, even if the source is faint, partiallyoccluded, or enveloped in the noisy and variable radiation background characteristic of urban scenes.In addition, BA provides simultaneous inference of source parameters such as the source intensity or source type while detecting it. The thesis demonstrates this capability and also develops techniques to efficiently optimize these parameters over large possible setting spaces. Methods developed in this thesis are demonstrated both in simulation and in a radiation-sensing backpack that applies robotic localization techniques to enable indoor surveillance of radiation sources.The thesis further improves the BA algorithm's capability to be robust under various detection scenarios. First, we augment BA with appropriate statistical models to improve estimation of signal components in low photon count detection, where the sensor may receive limited photon counts from either source and/or background. Second, we develop methods for online sensor reliability monitoring to create algorithms that are resilient to possible sensor faults in a data pipeline containing one or multiple sensors.Finally, we develop Retrospective BA, a variant of BA that allows reinterpretation of past sensor data in light of new information about percepts. These Retrospective capabilities include the use of Hidden Markov Models in BA to allow automatic correction of a sensor pipeline when sensor malfunction may be occur, an AnomalyMatch search strategy to efficiently optimize source hypotheses, and prototyping of a Multi-Modal Augmented PCA to more flexibly model background and nuisance source fluctuations in a dynamic environment.To my parents and cute little brothers and sisters too who keep copying everything I do and to our puppy of new Chewbie Doobie Doo ...And also to my iCreate robots.These iCreates are ones I worked on at University of Southern California (USC) in undergrad. After I graduated, (unknown to me at the time) several were shipped to Carnegie Mellon University (CMU) for use here. I found them in storage gathering dust. Their model is obsolete, but they still boot up, so I updated their PlayerStage c...

An Ensemble Approach to Computationally Efficient Radiological Anomaly Detection and Isotope Identification

“…2) Spectral anomaly detection (SAD): SAD projects measured spectra onto an PCA-based orthonormal subspace, then reconstructs the spectra using the orthonormal components. Then, the anomaly metric is obtained by measuring the reconstruction error [26], [27]. Suppose the orthonormal subspace is denoted as U.…”

Field Response Prediction: Framing the problem.

“…A similar background estimation technique is described in [44] and [45] under the name Energy Windowing Regression. In their case, the windows selection is based on match filtering to a source template in order to select a range of energies (called the source window ) likely containing the source signal.…”

“…By linearly regressing the background gamma counts of training data outside the source window (call those energy bins ), the number of background gamma counts inside the source window can be predicted. In [44], Least Squares estimator and Ridge Regression estimator are also presented.…”

“…In [45] (and more extensively in [44]), another method is presented to account for expected variations in background and common potential nuisance sources causing false alarms in urban searches with mobile spectrometers. Named Bayesian Aggregation (BA), this method relies on field data and injected synthetic sources to learn statistical models of expected threats.…”

Canonical Autocorrelation Analysis

“…In the radiation analysis domain, PCA-based anomaly detection is commonly used (Tandon 2015). PCA-based spectral anomaly detection works by essentially calculating the magnitude of the residual after a background-subtracting "projection", where the projection is either a strict projection onto the subspace spanned by the top few principal components of the covariance matrix, or a dilation modified projection where the correlation (not covariance) matrix is used to learn the low dimensional projection and then appropriate scaling of the measurement dimensions is performed before projection and scaled back after projection.…”