Previous studies have shown that experimentally implemented formant perturbations result in production of compensatory responses in the opposite direction of the perturbations. In this study, we investigated how participants adapt to a) auditory perturbations that shift formants to a specific point in the vowel space and hence remove variability of formants (focused perturbations), and b) auditory perturbations that preserve the natural variability of formants (uniform perturbations). We examined whether the degree of adaptation to focused perturbations was different from adaptation to uniform adaptations. We found that adaptation magnitude of the first formant (F1) was smaller in response to focused perturbations. However, F1 adaptation was initially moved in the same direction as the perturbation, and after several trials the F1 adaptation changed its course toward the opposite direction of the perturbation. We also found that adaptation of the second formant (F2) was smaller in response to focused perturbations than F2 responses to uniform perturbations. Overall, these results suggest that formant variability is an important component of speech, and that our central nervous system takes into account such variability to produce more accurate speech output.

Our ability to understand networks is important to many applications, from the analysis and modeling of biological networks to analyzing social networks. Unveiling network dynamics allows us to make predictions and decisions. Moreover, network dynamics models have inspired new ideas for computational methods involving multi-agent cooperation, offering effective solutions for optimization tasks. This dissertation presents new theoretical results on network inference and multi-agent optimization, split into two parts -

The first part deals with modeling and identification of network dynamics. I study two types of network dynamics arising from social and gene networks. Based on the network dynamics, the proposed network identification method works like a `network RADAR', meaning that interaction strengths between agents are inferred by injecting `signal' into the network and observing the resultant reverberation. In social networks, this is accomplished by stubborn agents whose opinions do not change throughout a discussion. In gene networks, genes are suppressed to create desired perturbations. The steady-states under these perturbations are characterized. In contrast to the common assumption of full rank input, I take a laxer assumption where low-rank input is used, to better model the empirical network data. Importantly, a network is proven to be identifiable from low rank data of rank that grows proportional to the network's sparsity. The proposed method is applied to synthetic and empirical data, and is shown to offer superior performance compared to prior work. The second part is concerned with algorithms on networks. I develop three consensus-based algorithms for multi-agent optimization. The first method is a decentralized Frank-Wolfe (DeFW) algorithm. The main advantage of DeFW lies on its projection-free nature, where we can replace the costly projection step in traditional algorithms by a low-cost linear optimization step. I prove the convergence rates of DeFW for convex and non-convex problems. I also develop two consensus-based alternating optimization algorithms --- one for least square problems and one for non-convex problems. These algorithms exploit the problem structure for faster convergence and their efficacy is demonstrated by numerical simulations.

I conclude this dissertation by describing future research directions.

Many mysteries still surround brain function, and yet greater understanding of it is vital to advancing scientific research. Studies on the brain in particular play a huge role in the medical field as analysis can lead to proper diagnosis of patients and to anticipatory treatments. The objective of this research was to apply signal processing techniques on electroencephalogram (EEG) data in order to extract features for which to quantify an activity performed or a response to stimuli. The responses by the brain were shown in eigenspectrum plots in combination with time-frequency plots for each of the sensors to provide both spatial and temporal frequency analysis. Through this method, it was revealed how the brain responds to various stimuli not typically used in current research. Future applications might include testing similar stimuli on patients with neurological diseases to gain further insight into their condition.

Divergence functions are both highly useful and fundamental to many areas in information theory and machine learning, but require either parametric approaches or prior knowledge of labels on the full data set. This paper presents a method to estimate the divergence between two data sets in the absence of fully labeled data. This semi-labeled case is common in many domains where labeling data by hand is expensive or time-consuming, or wherever large data sets are present. The theory derived in this paper is demonstrated on a simulated example, and then applied to a feature selection and classification problem from pathological speech analysis.

This work details the bootstrap estimation of a nonparametric information divergence measure, the Dp divergence measure, using a power law model. To address the challenge posed by computing accurate divergence estimates given finite size data, the bootstrap approach is used in conjunction with a power law curve to calculate an asymptotic value of the divergence estimator. Monte Carlo estimates of Dp are found for increasing values of sample size, and a power law fit is used to relate the divergence estimates as a function of sample size. The fit is also used to generate a confidence interval for the estimate to characterize the quality of the estimate. We compare the performance of this method with the other estimation methods. The calculated divergence is applied to the binary classification problem. Using the inherent relation between divergence measures and classification error rate, an analysis of the Bayes error rate of several data sets is conducted using the asymptotic divergence estimate.

Telemedicine is a multipurpose tool that allows medical professionals to use technology as a means to evaluate, diagnose, and treat patients remotely. This paper focuses on the challenges that developing telemedicine programs face, specifically discussing target population, user experience, and physician adoption. Various users of telemedicine share their experiences overcoming such challenges with the greater goal of this paper being to facilitate the growth of telemedicine programs.

Passive radar can be used to reduce the demand for radio frequency spectrum bandwidth. This paper will explain how a MATLAB simulation tool was developed to analyze the feasibility of using passive radar with digitally modulated communication signals. The first stage of the simulation creates a binary phase-shift keying (BPSK) signal, quadrature phase-shift keying (QPSK) signal, or digital terrestrial television (DTTV) signal. A scenario is then created using user defined parameters that simulates reception of the original signal on two different channels, a reference channel and a surveillance channel. The signal on the surveillance channel is delayed and Doppler shifted according to a point target scattering profile. An ambiguity function detector is implemented to identify the time delays and Doppler shifts associated with reflections off of the targets created. The results of an example are included in this report to demonstrate the simulation capabilities.

Epilepsy affects numerous people around the world and is characterized by recurring seizures, prompting the ability to predict them so precautionary measures may be employed. One promising algorithm extracts spatiotemporal correlation based features from intracranial electroencephalography signals for use with support vector machines. The robustness of this methodology is tested through a sensitivity analysis. Doing so also provides insight about how to construct more effective feature vectors.

The ability of cochlear implants (CI) to restore auditory function has advanced significantly in the past decade. Approximately 96,000 people in the United States benefit from these devices, which by the generation and transmission of electrical impulses, enable the brain to perceive sound. But due to the predominantly Western cochlear implant market, current CI characterization primarily focuses on improving the quality of American English. Only recently has research begun to evaluate CI performance using other languages such as Mandarin Chinese, which rely on distinct spectral characteristics not present in English. Mandarin, a tonal language utilizes four, distinct pitch patterns, which when voiced a syllable, conveys different meanings for the same word. This presents a challenge to hearing research as spectral, or frequency based information like pitch is readily acknowledged to be significantly reduced by CI processing algorithms. Thus the present study sought to identify the intelligibility differences for English and Mandarin when processed using current CI strategies. The objective of the study was to pinpoint any notable discrepancies in speech recognition, using voice-coded (vocoded) audio that simulates a CI generated stimuli. This approach allowed 12 normal hearing English speakers, and 9 normal hearing Mandarin listeners to participate in the experiment. The number of frequency channels available and the carrier type of excitation were varied in order to compare their effects on two cases of Mandarin intelligibility: Case 1) word recognition and Case 2) combined word and tone recognition. The results indicated a statistically significant difference between English and Mandarin intelligibility for Condition 1 (8Ch-Sinewave Carrier, p=0.022) given Case 1 and Condition 1 (8Ch-Sinewave Carrier, p=0.001) and Condition 3 (16Ch-Sinewave Carrier, p=0.001) given Case 2. The data suggests that the nature of the carrier type does have an effect on tonal language intelligibility and warrants further research as a design consideration for future cochlear implants.

Background: The majority of individuals with post-traumatic headache have symptoms that are indistinguishable from migraine. The overlap in symptoms amongst these individuals raises the question as to whether post-traumatic headache has a unique pathophysiology or if head trauma triggers migraine. The objective of this study was to compare brain structure in individuals with persistent post-traumatic headache (i.e. headache lasting at least 3 months following a traumatic brain injury) attributed to mild traumatic brain injury to that of individuals with migraine.

Methods: Twenty-eight individuals with persistent post-traumatic headache attributed to mild traumatic brain injury and 28 individuals with migraine underwent brain magnetic resonance imaging on a 3 T scanner. Regional volumes, cortical thickness, surface area and curvature measurements were calculated from T1-weighted sequences and compared between subject groups using ANCOVA. MRI data from 28 healthy control subjects were used to interpret the differences in brain structure between migraine and persistent post-traumatic headache.

Results: Differences in regional volumes, cortical thickness, surface area and brain curvature were identified when comparing the group of individuals with persistent post-traumatic headache to the group with migraine. Structure was different between groups for regions within the right lateral orbitofrontal lobe, left caudal middle frontal lobe, left superior frontal lobe, left precuneus and right supramarginal gyrus (p < .05). Considering these regions only, there were differences between individuals with persistent post-traumatic headache and healthy controls within the right lateral orbitofrontal lobe, right supramarginal gyrus, and left superior frontal lobe and no differences when comparing the migraine cohort to healthy controls.

Conclusions: In conclusion, persistent post-traumatic headache and migraine are associated with differences in brain structure, perhaps suggesting differences in their underlying pathophysiology. Additional studies are needed to further delineate similarities and differences in brain structure and function that are associated with post-traumatic headache and migraine and to determine their specificity for each of the headache types.