No description available.

No description available.

No description available.

No description available.

No description available.

The main focus of this research is to detect vulnerabilities on the Mississippi river levees using remotely sensed Synthetic Aperture Radar (SAR) imagery. Unstable slope conditions can lead to slump slides, which weaken the levees and increase the likelihood of failure during floods. On-site inspection of levees is expensive and time-consuming, so there is a need to develop efficient automated techniques based on remote sensing technologies to identify levees that are more vulnerable to failure under flood loading. Synthetic Aperture Radar technology, due to its high spatial resolution and potential soil penetration capability, is a good choice to identify problem areas along the levee so that they can be treated to avoid possible catastrophic failure. This research analyzes the ability of detecting the slump slides on the levee with different frequency bands of SAR data. The two SAR datasets used in this study are: (1) the L-band airborne radar data from NASA JPL’s Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR), and (2) the X-band satellite-based radar data from DLR’s TerraSAR-X (TSX). The main contribution of this research is the development of a machine learning framework to 1) provide improved knowledge of the status of the levees, 2) detect anomalies on the levee sections, and 3) provide early warning of impending levee failures. Polarimetric and textural features have been computed and utilized in the classification tasks to achieve efficient levee characterization. Various approaches of image analysis methods for characterizing levee segments within the study area have been implemented and tested. The RX anomaly detector, a trainingree unsupervised classification algorithm, detected the active slump slides on the levee at the time of image acquisition and also flagged some areas as “anomalous”, where new slides appeared at a later date. This technique is very fast and does not depend on ground truth information, so these results guide levee managers to investigate the areas shown as anomalies in the classification map. The support vector machine (SVM) supervised learning algorithm with grey level co-occurrence matrix (GLCM) features provided excellent results in identifying slump slides on the levee.

Time series are very common in presenting collected data such as economic indicators, natural phenomenon, control engineering data, among others. In the last decade, the interest in time series data mining increased as the amount of collected data increased dramatically. Standard approaches for time series classification are based on collecting distance measures, such as the Euclidian distance (ED) and dynamic time warping (DTW) along with 1-NN classifier for further classification. Recently, more advanced types of classification were found, introducing primitives (named time series shapelet) that consistently represent a certain class. The time series shapelet is a small sub-section of the entire time series, which is “particularly discriminating”. It appears that shapelets based classification produces higher accuracies on some data sets, based on the fact that the global features are more sensitive to noise than locals. Despite its advantages, the time series shapelets classification has an apparent disadvantage: very slow training time. This work attempts to improve the training time for the originally proposed time series shapelets classification algorithm and introduces a new approach for time series classification based on concatenated decision tree paths. First, the classical algorithm for time series classification based on shapelets, is significantly improved in terms of the training time. The improvement is based on using randomly generated sequences tuned in a particle-swarm-optimization (PSO) environment, instead of using sub-series from the original time series. Second, a new highly accurate classification method, based on concatenated decision tree paths, is introduced. The approach builds a unique representative pattern of a certain class based on the taken paths in a pool of decision trees. Third, the proposed method has been successfully extended for a 2-class-labels classification problem where only one decision tree can be built. A variety of 2-class-labels decision trees were built based on different splitting criterion (distance to a random shapelet); thus- increasing the pool of decision trees and increasing the overall accuracy. Fourth, the proposed method was successfully applied on two classes image classification problem, by converting the image into time series. An accuracy of around 95% was achieved for the pedestrian detection case from the Daimler database.

Bayesian networks are a widely used graphical model which formalize reasoning under uncertainty. Unfortunately, construction of a Bayesian network by an expert is timeconsuming, and, in some cases, all expertsmay not agree on the best structure for a problem domain. Additionally, for some complex systems such as those present in molecular biology, experts with an understanding of the entire domain and how individual components interact may not exist. In these cases, we must learn the network structure from available data. This dissertation focuses on score-based structure learning. In this context, a scoring function is used to measure the goodness of fit of a structure to data. The goal is to find the structure which optimizes the scoring function. The first contribution of this dissertation is a shortest-path finding perspective for the problem of learning optimal Bayesian network structures. This perspective builds on earlier dynamic programming strategies, but, as we show, offers much more flexibility. Second, we develop a set of data structures to improve the efficiency of many of the integral calculations for structure learning. Most of these data structures benefit our algorithms, dynamic programming and other formulations of the structure learning problem. Next, we introduce a suite of algorithms that leverage the new data structures and shortest-path finding perspective for structure learning. These algorithms take advantage of a number of new heuristic functions to ignore provably sub-optimal parts of the search space. They also exploit regularities in the search that previous approaches could not. All of the algorithms we present have their own advantages. Some minimize work in a provable sense; others use external memory such as hard disk to scale to datasets with more variables. Several of the algorithms quickly find solutions and improve them as long as they are given more resources. Our algorithms improve the state of the art in structure learning by running faster, using less memory and incorporating other desirable characteristics, such as anytime behavior. We also pose unanswered questions to drive research into the future.

Voltage source inverters have become widely used in the last decade primarily due to the fact that the dangers and limitations of relying on fossil fuel based power generation have been seen and the long term effects felt especially with regards to climate change. Policies and targets have been implemented such as from the United Nations climate change conference (COPxx) concerning human activities that contribute to global warming from individual countries. The most effective way of reducing these greenhouse gases is to turn to renewable energy sources such as the solar, wind etc instead of coal. Converters play the crucial role of converting the renewable source dc power to ac single phase or multiphase. The advancement in research in renewable energy sources and energy storage has made it possible to do things more efficiently than ever before. Regular or 2 level inverters are adequate for low power low voltage applications but have drawbacks when being used in high power high voltage applications as switching components have to be rated upwards and also switch between very high potential differences. To lessen the constraints on the switching components and to reduce the filtering requirements, multilevel inverters (MLI's) are preferred over two level voltage source inverters (VSI's). This thesis discusses the implementation of various types of MLI's and compares four different pulse width modulation (pwm) techniques that are often used in MLI under consideration: three, five, seven and nine level inverters. Harmonic content of the output voltage is recorded across a range of modulation indices for each of the three popular topologies in literature. Output from the inverter is filtered using an L only and an LC filter whose design techniques are presented. A generalized prediction algorithm using machine learning techniques to give the value of the expected THD as the modulation index is varied for a specific topology and PWM switching method is proposed in this study. Simulation and experimental results are produced in five level form to verify and validate the proposed algorithm.

In almost all computer vision and perception based applications, particularly with camera and lidar; state-of-the-art algorithms are all based upon deep neural networks which require large amounts of data. Thus, the ability to label data accurately and quickly is of great importance. Approaches to semi-automated labeling (SAL) thus far have relied on using state-of-the-art object detectors to assist with labeling; however, these approaches still require a significant number of manual corrections. Surprisingly, none of these approaches have considered labeling from the perspective of multiple diverse algorithms. In this thesis a new framework for semi-automated labeling is presented, it is called F-SAL which stands for Fusion Based Semi-automated Labeling. Firstly, F-SAL extends on the idea of SAL through introducing multi-algorithm fusion with learning based feedback. Secondly, it incorporates new stages such as uncertainty evaluation and diversity evaluation. All the algorithms and design choices regarding localization fusion, label fusion, uncertainty and diversity evaluation are presented and discussed in significant detail. The biggest advantage of F-SAL is that through the fusion of algorithms, the number of true detections is either more or equivalent to the best single detector; while the false alarms are suppressed significantly. In the case of a single detector, to lower the false alarm rate, detector parameters must be adjusted, which trade lower false alarms for fewer detections. With F-SAL, a lower false alarm rate can be achieved without sacrificing any detections, as false alarms are suppressed during fusion, and true detections are maximized through diversity. Results on several datasets for image and lidar data show that F-SAL outperforms the single best detector in all scenarios. / Thesis / Master of Applied Science (MASc)