I studied translocation of Texas horned lizards on Tinker Air Force Base, Midwest City, Oklahoma, using correlative and mechanistic habitat suitability models. My goals were broadly classified into two categories: first, to determine if the addition of mechanistic data layers (i.e., habitat-niche models) in a correlative model improved the overall accuracy of model predictions, and second, to apply the best model produced from my dataset to a planned translocation event on Tinker Air Force Base. Correlative data layers (i.e., habitat models) included typically applied datasets such as vegetative components, Euclidean distance statistics, neighborhood analyses, and topographically-derived information. Mechanistic data layers were estimates of thermal suitability derived from field-collected datasets and biophysical calculations, and estimates of prey availability taken from interpolated datasets. I estimated habitat suitability using the partitioned Mahalanobis distance statistic, which is a suitable model technique for presence-only data. Translocated and resident lizards were monitored via radiotelemetry and using fluorescent powder trails. Telemetry locations and powder trails were overlaid onto habitat suitability models to provide the datasets used to quantify interaction between site occupancy and habitat model predictions. Lizard paths were tested against random walk models to determine efficiency of travel, and site occupancy metrics (powder track and telemetry Mahalanobis distance values) were tested using parametric (repeated-measures ANOVA) and nonparametric (Wilcoxon rank-sum and signed-rank tests) tests. Mechanistic data layers did not substantially improve model accuracy over correlative-only layers, and data layers taken from mixed bare soil-vegetation, shrub, and grassland habitat types dominated important eigenvector weights. Analyses of fluorescent powder track data suggested that lizards did not move through habitat differently from a random walk model, potentially due to neighborhood factor loadings strongly influencing the area in which entire trails traveled. Wilcoxon tests and repeated-measures ANOVA results suggested that although lizards experienced different median Mahalanobis distance values by group (translocated, resident), there appeared to be an overall decrease in distance scores for translocated individuals over time. In this context, translocated individuals seemed to acclimate their behavior to areas that were predicted to be more suitable by Mahalanobis classifiers. Although survival results were not encouraging and habitat models did not suggest that my translocation site was ideal, my data supports the idea that translocations may be aided in the future by modeling efforts. My models suggest that mechanistic data layers may not improve classification accuracy over correlative processes, but this may be due to inaccurate representation of specific mechanisms over spatial and temporal scales. Future work should focus on including more explicit measures of mechanisms, as well as broadening biotic influences on species distributions (i.e., predator distribution, intra- and interspecific competition).

EMG classification is widely used in electric control of mechanically developed prosthesis, robots development, clinical application etc. It has been evaluated for years, but the main goal of this research is to develop an easy to implement and fast to execute pattern recognition method for classifying signals used for human gait analysis. This method is based on adding two new temporal features (form factor and standard deviation) for EMG signal recognition and using them along with several popular features (area under the curve, wavelength function-pathway and zero crossing rate) to come up with a low complexity suitable feature extraction. Results are presented for EMG data and a comparison with existing methods is made to validate the applicability of the foregoing method. It is shown that the best combination in terms of accuracy and time performance is given by spectral and temporal extraction features along with neural network recognition (NN) algorithm.

Poster exhibited at GPSC Student Showcase, February 24th, 2016, University of Arizona. / Motivation: The conventional approach to personalized medicine relies on molecular data analytics across multiple patients. The path to precision medicine lies with molecular data analytics that can discover interpretable single-subject signals (N-of-1). We developed a global framework, N-of-1-pathways, for a mechanistic-anchored approach to single-subject gene expression data analysis. We previously employed a metric that could prioritize the statistical significance of a deregulated pathway in single subjects, however, it lacked in quantitative interpretability (e.g. the equivalent to a gene expression fold-change). Results: In this study, we extend our previous approach with the application of statistical Mahalanobis distance (MD) to quantify personal pathway-level deregulation. We demonstrate that this approach, N-of-1-pathways Paired Samples MD (N-OF-1-PATHWAYS-MD), detects deregulated pathways (empirical simulations), while not inflating false-positive rate using a study with biological replicates. Finally, we establish that N-OF-1-PATHWAYS-MD scores are, biologically significant, clinically relevant and are predictive of breast cancer survival (P<0.05, n¼80 invasive car- cinoma; TCGA RNA-sequences). Conclusion: N-of-1-pathways MD provides a practical approach towards precision medicine. The method generates the magnitude and the biological significance of personal deregulated pathways results derived solely from the patient’s transcriptome. These pathways offer the opportunities for deriving clinically actionable decisions that have the potential to complement the clinical interpret- ability of personal polymorphisms obtained from DNA acquired or inherited polymorphisms and mutations. In addition, it offers an opportunity for applicability to diseases in which DNA changes may not be relevant, and thus expand the ‘interpretable ‘omics’ of single subjects (e.g. personalome).

In this thesis, we consider the clustering of time series data; specifically, time series that can be modeled in the state space framework. Of primary focus is the pairwise discrepancy between two state space time series. The state space model can be formulated in terms of two equations: the state equation, based on a latent process, and the observation equation. Because the unobserved state process is often of interest, we develop discrepancy measures based on the estimated version of the state process. We compare these measures to discrepancies based on the observed data. In all, seven novel discrepancies are formulated. First, discrepancies derived from Kullback-Leibler (KL) information and Mahalanobis distance (MD) measures are proposed based on the observed data. Next, KL information and MD discrepancies are formulated based on the composite marginal contributions of the smoothed estimates of the unobserved state process. Furthermore, an MD is created based on the joint contributions of the collection of smoothed estimates of the unobserved state process. The cross trajectory distance, a discrepancy heavily influenced by both observed and smoothed data, is proposed as well as a Euclidean distance based on the smoothed state estimates. The performance of these seven novel discrepancies is compared to the often used Euclidean distance based on the observed data, as well as a KL information discrepancy based on the joint contributions of the collection of smoothed state estimates (Bengtsson and Cavanaugh, 2008). We find that those discrepancy measures based on the smoothed estimates of the unobserved state process outperform those discrepancy measures based on the observed data. The best performance was achieved by the discrepancies founded upon the joint contributions of the collection of unobserved states, followed by the discrepancies derived from the marginal contributions. We observed a non-trivial degradation in clustering performance when estimating the parameters of the state space model. To improve estimation, we propose an iterative estimation and clustering routine based on the notion of finding a series' most similar counterparts, pooling them, and estimating a new set of parameters. Under ideal circumstances, we show that the iterative estimation and clustering algorithm can potentially achieve results that approach those obtained in settings where parameters are known. In practice, the algorithm often improves the performance of the model-based clustering measures. We apply our methods to two examples. The first application pertains to the clustering of time course genetic data. We use data from Cho et al. (1998) where a time course experiment of yeast gene expression was performed in order to study the yeast mitotic cell cycle. We attempt to discover the phase to which 219 genes belong. The second application seeks to answer whether or not influenza and pneumonia mortality can be explained geographically. Data from a collection of cities across the U.S. are acquired from the Morbidity and Mortality Weekly Report (MMWR). We cluster the MMWR data without geographic constraints, and compare the results to clusters defined by MMWR geographic regions. We find that influenza and pneumonia mortality cannot be explained by geography.

Person re-identification (Re-ID) has become an intense research area in recent years. The main goal of this topic is to recognize and match individuals over time at the same or different locations. This task is challenging due to the variation of illumination, viewpoints, pedestrians’ appearance and partial occlusion. Previous works mainly focus on finding robust features and metric learning. Many metric learning methods convert the Re-ID problem to a matrix decomposition problem by Fisher discriminant analysis (FDA). Mahalanobis distance metric learning is a popular method to measure similarity; however, since directly extracted descriptors usually have high dimensionality, it’s intractable to learn a high-dimensional semi-positive definite (SPD) matrix. Dimensionality reduction is used to project high-dimensional descriptors to a lower-dimensional space while preserving those discriminative information. In this paper, the kernel Fisher discriminant analysis (KLFDA) [38] is used to reduce dimensionality given that kernelization method can greatly improve Re-ID performance for nonlinearity. Inspired by [47], an SPD matrix is then learned on lower-dimensional descriptors based on the limitation that the maximum intraclass distance is at least one unit smaller than the minimum interclass distance. This method is proved to have excellent performance compared with other advanced metric learning.

Master of Science / Department of Electrical and Computer Engineering / Christopher L. Lewis / This thesis describes the development of an automated computer vision system that identifies and inventories road signs from imagery acquired from the Kansas Department of Transportation's road profiling system that takes images every 26.4 feet on highways through out the state. Statistical models characterizing the typical size, color, and physical location of signs are used to help identify signs from the imagery. First, two phases of a computationally efficient K-Means clustering algorithm are applied to the images to achieve over-segmentation. The novel second phase ensures over-segmentation without excessive computation. Extremely large and very small segments are rejected. The remaining segments are then classified based on color. Finally, the frame to frame trajectories of sign colored segments are analyzed using triangulation and Bundle adjustment to determine their physical location relative to the road video log system. Objects having the appropriate color, and physical placement are entered into a sign database. To develop the statistical models used for classification, a representative set of images was segmented and manually labeled determining the joint probabilistic models characterizing the color and location typical to that of road signs. Receiver Operating Characteristic curves were generated and analyzed to adjust the thresholds for the class identification. This system was tested and its performance characteristics are presented.

In the event of a natural or anthropogenic disturbance, environmental resource managers require a reliable tool to quickly assess the spatial extent of potential damage to the seagrass resource. The temporal availability of the Landsat 5 Thematic Mapper (TM) imagery, 16-20 days, provides a suitable option to detect and assess damage to the seagrass resource. In this study, remote sensing Landsat 5 TM imagery is used to map the spatial extent of the seagrass resource. Various classification techniques are applied to delineate the seagrass beds in Clearwater Harbor and St. Joseph Sound, FL. This study aims to determine the most appropriate seagrass habitat mapping technique by evaluating the accuracy and validity of the resultant classification maps. Field survey data and high resolution aerial photography are available to use as ground truth information. Seagrass habitat in the study area consists of seagrass species and rhizophytic algae; thus, the species assemblage is categorized as submerged aquatic vegetation (SAV). Two supervised classification techniques, Maximum Likelihood and Mahalanobis Distance, are applied to extract the thematic features from the Landsat imagery. The Mahalanobis Distance classification (MDC) method achieves the highest overall accuracy (86%) and validation accuracy (68%) for the delineation of the presence/absence of SAV. The Maximum Likelihood classification (MLC) method achieves the highest overall accuracy (74%) and validation accuracy (70%) for the delineation of the estimated coverage of SAV for the classes of continuous and patchy seagrass habitat. The soft classification techniques, linear spectral unmixing (LSU) and artificial neural network (ANN), did not produce reasonable results for this particular study. The comparison of the MDC and MLC to the current Seagrass Aerial Photointerpretation (AP) project indicates that the classification of SAV from Landsat 5 TM imagery provides a map product with similar accuracy to the AP maps. These results support the application of remote sensing thematic feature extraction methods to analyze the spatial extent of the seagrass resource. While the remote sensing thematic feature extraction methods from Landsat 5 TM imagery are deemed adequate, the use of hyperspectral imagery and better spectral libraries may improve the identification and mapping accuracy of the seagrass resource.

In this thesis, we describe two approaches to automatically synthesize the emotional speech of a target speaker based on the hidden Markov model for his/her neutral speech. In the interpolation based method, the basic idea is the model interpolation between the neutral model of the target speaker and an emotional model selected from a candidate pool. Both the interpolation model selection and the interpolation weight computation are determined based on a model-distance measure. We propose a monophone-based Mahalanobis distance (MBMD). In the parallel model combination (PMC) based method, our basic idea is to model the mismatch between neutral model and emotional model. We train linear regression model to describe this mismatch. And then we combine the target speaker neutral model with the linear regression model. We evaluate our approach on the synthesized emotional speech of angriness, happiness, and sadness with several subjective tests. Experimental results show that the implemented system is able to synthesize speech with emotional expressiveness of the target speaker.

Non-invasive physical examination helps to make disease diagnosis with minimum injury to the body. Cardiology ultrasound is a non-invasive examination which can be used as a auxiliary tool for diagnose cardiac structure abnormalities. With more understanding of heart diseases, it has been recognized that heart failures are closely related to left ventricular systolic and diastolic function. Following Chen (2011) and Kao (2011), we study association of heart diseases with the change of gray-scale values in the cardiology ultrasound images of left ventricular systolic and diastolic. Since data obtained from ultrasound image is of matrix type with high dimensions, following the method proposed by Chen (2011) and Kao (2011), factor scores obtained from factor analysis are used as a basis for classification. We take the factor scores of normal subjects to establish the bench mark and calculate the Mahalanobis distance of each abnormal subject with the model established by the data from normal group. Later based on this distance to the normal group, cardiac function of the subject is distinguished as normal or not. In order to improve the accuracy of the classification, influential points which may cause inaccurate covariance matrix estimate on the subjects in normal group are identified. Based on concepts from optimal designs theory, some criteria are established for screening out the influential points.

¡@¡@In this thesis, we use a hidden Markov model which can use a small amount of corpus to synthesize speech with certain quality to implement speech synthesis system for Chinese. More, the emotional speech are synthesized by the flexibility of the parametric speech in this model. We conduct model interpolation and model adaptation to synthesize speech from neutral to particular emotion without target speaker¡¦s emotional speech. In model adaptation, we use monophone-based Mahalanobis distance to select emotional models which are close to target speaker from pool of speakers, and estimate the interpolation weight to synthesize emotional speech. In model adaptation, we collect abundant of data training average voice models for each individual emotion. These models are adapted to specific emotional models of target speaker by CMLLR method. In addition, we design the Latin-square evaluation to reduce the systematic offset in the subjective tests, making results more credible and fair. We synthesize emotional speech include happiness, anger, sadness, and use Latin square design to evaluate performance in three part similarity, naturalness, and emotional expression respectively. According to result, we make a comprehensive comparison and conclusions of two method in emotional speech synthesis.