Search

Multi-view multiple people detection, labelling and tracking has attracted considerable interest in recent years and has been employed in many application domains such as video surveillance, human pose estimation and 3D human modelling. However, correctly detecting, labelling and tracking multiple people from different viewpoints is a challenging task, especially in complex scenes, due to occlusions, visual ambiguities, as well as variations in appearance and illumination. Recent approaches established based on deep learning such as object detection and semantic segmentation have proved very successful at improving the performance of a wide range of computer vision tasks. By contrast, limited works have been done with deep learning in multi-view multiple people detection, labelling and tracking, due to the lack of multi-view datasets. This thesis therefore introduces novel algorithms to advance the state-of-the-art in multi-view detection, labelling and tracking on both images and videos based on deep learning, whilst also establishing new synthetic datasets to address the lack of multi-view data in this field. Firstly, we propose a multi-view framework for joint object detection and labelling based on pairs of images. As the lack of multi-view data prevents training an end-to-end network for the task, we therefore leverage some of the benefits of deep learning by introducing a novel singleview convolutional neural network (CNN) with a series of multi-view constraints. The proposed framework extends the single-view Mask R-CNN approach to multiple views to achieve multiview object detection and labelling without need for additional training, thus overcoming the lack of multi-view data. Dedicated components are embedded into the framework to match objects across views by enforcing epipolar constraints, appearance feature similarity and class coherence. By jointly processing detection and labelling in a unified network, the multi-view extension enables the proposed framework to detect objects which would otherwise be mis-detected in a single view, and achieves coherent object labelling across views. Next, to fundamentally address the shortage of multi-view dataset and leverage the advantage of deep neural network directly into multi-view detection and labelling, we generate a largescale synthetic dataset named MV3DHumans for the multi-view multiple people detection and labelling task by combining 3D human models, panoramic backgrounds, human poses, and appearance detail rendering. The proposed MV3DHumans dataset consists of various scenes with different degrees of crowdedness, providing sufficient data for training deep neural networks. We then develop a novel end-to-end multi-view framework named Multi-View Labelling network (MVL-net), using a CNN to learn multi-view consistency based on the generated synthetic dataset. A novel labelling branch with a matching net is introduced in the MVL-net to predict matching confidence scores for pairs of people from two views, thus solving the problem of label prediction for an unknown number of people. In addition, a calibration component based on the epipolar geometry is integrated in the proposed framework to further improve the performace. Finally, we extend our work from multi-view labelling on images to the more challenging tracking problem on multi-view videos. A synthetic multi-view video dataset named MV3DHumansVideo is established by extending data generation from images to videos, consisting of hundreds of short multi-view video sequences for training and several long sequences for testing. Based on the newly proposed multi-view video dataset, we propose a novel spatiotemporal correlation (STC) network, jointly achieving detection, temporal feature learning and spatial feature learning based on multi-view videos. The proposed STC network combines detection and feature learning in a unified network and seamlessly integrates temporal and spatial context information to enhance temporal feature and spatial feature learned for multi-view video tracking. An efficient multi-view tracking framework is then developed based on learned features to obtain consistent people labels for multi-view videos. We demonstrate that this method can learn effective features for multi-view tracking and obtain consistent people labels for entire multi-view videos.

There is a need to address disparities in the underrepresented minority (URM) STEM students' degree attainment and retention, vs. non-URM students. A literature review suggests that URM STEM students face barriers related to demographic, academic, and social-cognitive factors. The Louis Stokes North Star STEM Alliance (LS-NSSA) seeks to address these factors and promote URM student success in STEM disciplines. However, there is a gap in knowledge regarding LS-NSSA's mechanism and outcomes. The current study investigates factors influencing URM STEM students' retention and graduation, investigates LS-NSSA program effectiveness, and investigates the path by which outcomes are achieved. A theory-based evaluation approach is utilized in an explanatory sequential mixed-methods design guided by LS-NSSA's theory of change. The results of three sub-studies suggest the importance of first-semester experience and academic outcomes for URM STEM students' graduation and retention. Participation in LS-NSSA is associated with a higher level of academic preparation to pursue a bachelor's degree, compared to URM students not affiliated with the program. URM STEM students participating in LS-NSSA's research mentorship programs have higher levels of confidence, interest, science identity, sense of belonging, and commitment to STEM. The current study offers evidence supporting LS-NSSA's theory of change regarding student academic and social-cognitive trajectories. The current study suggests directions for future evaluative studies of LS-NSSA. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

In recent years, polymeric products of micro-/nano dimensions have attracted considerable attention, and complex structures have been demonstrated to be generated from solid and encapsulated particles/fibres. Although various approaches have been developed to make these products, the bioactive agents are poorly controlled and bio-stable due to the drawbacks such as high costs of processing and the requirement for non-biocompatible solvents and additives. In contrast, electrohydrodynamic atomization (EHDA) and co-axial EHDA (CEHDA) processes are simple and has been extensively studied as a popular technology for producing and processing micro-/nanostructured materials containing bioactive molecules. The simplicity, flexibility of EHDA experimental setup, and efficiency of generating particulate materials at the micro-/nanoscale, with controllable size, shape, and morphology, make electrospraying to become one of the most fascinating tools to be employed in pharmaceutics, food, and healthcare fields. The first objective of this research was to see how different physical and mechanical processing parameters affected the encapsulated particles manufactured by the EHDA method. The second objective of this work was to encapsulate progesterone into polymer matrix using single needle EHDA system, and further improving aqueous solubility and sustained release of progesterone by varying Poly (D,L lactide-co-glicolide acid) (PLGA) copolymer composition and polymer:drug ratio. The third objective of this study was to use a CEHDA system to prepare and investigate the release of progesterone from core-shell nanoparticles. The fourth objective was to use an CEHDA system incorporating a tri-axial capillary device to fabricate multidrug (saxagliptin and dapagliflozin) loaded triple-layered biodegradable nanoparticles for the potential treatment of type 2 diabetes. The last objective was to prepare multidrug loaded double-layered and uniform structured nanoparticles and compare the drug release profiles between different particle configurations. This work reveals a powerful approach for producing micro and nano drug-loaded polymeric particles with modified release behaviour and improved physiological solubility and bioavailability of poorly water-soluble drugs by fulfilling these goals.

With $49 trillion of assets under management and thousands of funds in the market, the mutual fund industry around the world has been one of the biggest and most important financial intermediaries. Whether fund-families offer high-quality products to their investors affects the wealth, and in particular the after-retirement wealth, of millions of households. Protecting investors' interests requires weeding out worse performing funds. In the light of this, this thesis aims to investigate market clearing mechanisms, i.e. mutual fund exit decisions in the U.S. and the U.K markets. There is no reason to assume that fund-families' exit decisions are time invariant. This thesis argues that while fund-families had sufficient motivations to close down or re-organise the worse performing funds during non-financial-crisis times, this may not be so during the 2008 financial crisis. The distortion of the market clearing mechanisms during the financial crisis existed in both the U.S. and in the U.K. market. The thesis finds some evidence that funds' exit decisions were affected by directors' own remuneration and career concerns and that this was more visible during the financial crisis than during the other times. Industry structure can affect the strength of market forces to exit low-quality products. As the asset allocation of US funds is not closely monitored or highly regulated, fund managers are free to 'game' their investment objectives, which makes it hard to separate the whole U.S. market to create meaningful and distinguished sectors suitable for industry structure analysis. Fortunately, the U.K. mutual fund industry is suitable for such a study. In the UK, the asset allocation of a fund is closely monitored by the Investment Association (IA) to make sure that funds make investments according to their official investment objectives. This makes the actual asset allocation of a fund reflect well the fund's official (declared) investment objectives. Thus, it is more meaningful to separate the UK market into different sectors according to funds' declared investment objectives in comparison with the US market. The empirical analysis for each sector in the U.K. market indicates that the market clearing mechanisms were only effective in competitive sectors. The thesis contributes to the literature on the determinants of mutual fund exits by arguing that the role of poor performance in determining a fund's fate weakened during the financial crisis. It deepens the understanding of the role of fund boards in protecting shareholders' interests in different times. It fills the gap in the U.K. mutual fund research by comprehensively investigating, for the first time, the determinants of fund exits over time. Finally, the thesis adds to the organisation studies literature by confirming the importance of industry structure (competitiveness) in exiting poorly performing funds out of the market.

Over the last few decades, metal matrix syntactic foams (MMSF) have received a large amount of interest in both academia and industry, due to lightweight, high specific stiffness, strength and energy absorption capability. Their unique properties and multifunctionality allow them to be used in many different applications throughout different industrial sectors. This thesis investigated the compressive, energy absorption and energy dissipation properties of Al matrix syntactic foams. Ceramic microspheres (CM) with three different particle size ranges (75-150 μm, 125-250 μm and 250-500 μm) and the Al 6082 alloy were used to fabricate MMSFs. The Al/CM syntactic foams with different CMs and varying Al/CM ratios were fabricated by the pressure infiltration casting method. The compressive and energy absorption properties of all the types of syntactic foams were measured by static compression tests. The compressive strength of the syntactic foams was mainly determined by strength of both the Al matrix and the CM particles. The strength of the syntactic foams increased slightly with the volume percentage of CM particles. The densification strain of all the syntactic foams was dependent upon the porosity arising from the CM particles. The energy absorption of the syntactic foam was determined by both plateau strength and densification strain. The energy absorption of bimodal syntactic foams mixing large and small CM particles was nearly twice of the monomodal syntactic foams. Among the three types of the monomodal syntactic foams, the large CM particle reinforced syntactic foam absorbed the highest amount of energy under the Charpy impact. The energy dissipation properties of the Al/CM syntactic foams were assessed by intermittent compression tests. Hysteresis loops, which represent an energy dissipation, were present in the stress-strain curves of all the syntactic foams samples. The specific damping capacity of all syntactic foams was not affected by the CM particle size but was dependent upon the volume percentage of the CM particles in the syntactic foam. The dissipated energy was proportional to the square of the applied stress in the cyclic loading. One-stage repetitive cyclic compression tests were conducted on the monomodal syntactic foams and the syntactic foams toughened with Al particles. The specific damping capacity the syntactic foams evidentially decreased after 20 cycles. The hysteretic energy dissipation in repetitive cyclic compression of syntactic foams was sensitive to the CM particle size and the volume percentages of the CM particles. Two-stage repetitive cyclic compression tests were conducted on one monomodal syntactic foam. The cyclic loading history was found to have influence on the hysteretic energy dissipation. When the stress level was below the maximum stress experienced in the previous cycles, the specific dissipated energy and specific damping capacity were much lower than those measured at the same stress in the previous cycles. The hysteretic energy dissipation was caused by microcrack formation and propagation in the CM particles. The accumulation of crack development led to CM particle fracture, which resulted in a small permanent plastic deformation. The energy dissipation behaviour of syntactic foams with different sizes and volume percentages of the CM particles under impulsive loading was assessed. Both the particle size and volume percentage of the CM particles had an influence on the energy dissipation capacity of the Al/CM syntactic foams under free oscillation. As in cyclic compression, the damping of the syntactic foams in free oscillation was also caused by microcrack initiation and propagation in the CM particles.

Faithful replication of DNA and correct segregation of duplicated chromosomes into two daughter cells are essential to ensure genome integrity and cell survival. Genome integrity is constantly threatened by endogenous and exogenous sources of DNA damage. Consequently, eukaryotic cells have evolved the DNA damage response (DDR), a signalling network that monitors and repairs DNA lesions promptly and efficiently. The ATR-Chk1 pathway is an essential component of the DDR, which is activated by single-strand DNA (ssDNA) generated as a result of DNA replication stress in S phase. Chk1 is a threonine/serine kinase that transduces the DNA damage signal and promotes cell cycle arrest to allow time for DNA repair. The Chk1 structure consists of conserved N-terminal kinase (Chk1KD) and C-terminal regulatory domains (Chk1RD). Available evidence has shown that the isolated Chk1KD is constitutively active, but how the enzymatic activity is regulated in the context of the full length (Chk1FL) kinase is unknown. In this thesis, the relative enzyme efficiency of purified, recombinant Chk1KD and Chk1FL was quantitatively studied using kinase assays. It was found that the enzyme efficiency of Chk1FL was up to two orders of magnitude lower than that of Chk1KD. Biophysical and biochemical characterisation of Chk1FL provided evidence of a compact shape and an intramolecular association of the Chk1RD with the Chk1KD. A putative Chk1RD-binding site on the Chk1KD surface was identified and disrupted by mutagenesis, resulting in partial increase of Chk1FL kinase activity, thus supporting an inhibitory role of the Chk1RD in controlling kinase activity. Activation of Chk1 requires its phosphorylation by the PI3K-like ATR kinase and its recruitment to the replisome via a direct interaction with the core replisome component, Claspin. The Chk1-binding domain of Claspin contains a tandem repeat of three phosphothreonine/serine motifs, which need to be phosphorylated for interaction. However, the structural basis of Chk1 binding remains presently unclear. The Chk1-Claspin interaction was analysed by alanine scanning of Claspin's Chk1-binding motif and Chk1 binding to the mono-phosphorylated Chk1-binding domain of Claspin prepared using the amber codon suppression method. Furthermore, the affinity of Chk1KD and Chk1FL for a phosphorylated Claspin peptide spanning a single Chk1-binding motif or the three motifs was determined using fluorescence polarization and bio-layer interferometry. Chk1FL kinase activity was increased by the presence of a Claspin peptide corresponding to a single phosphorylated Chk1-binding motif. The results of my thesis provide new insights into the mechanism of Chk1 function. They support a model of Chk1 auto-inhibition mediated by the intramolecular interaction of its kinase and regulatory domains, and of Chk1 enhanced activity promoted by interaction with its replisome partner Claspin.

in vivo degradation of Ti implants has remained as a concern despite its perceived excellent corrosion resistance. Elevated levels of metal have been detected both locally and systemically, often leading to unfavourable biological responses. Corrosion of Ti, Zr and their alloys (Ti6Al4V and TiZr) were studied in simulated physiological solutions. It was found that albumin, the most prevalent tissue fluid protein, induced a time dependent dissolution of Ti6Al4V in the presence of H2O2, an inflammatory biomolecule commonly found in peri implant sites, in 0.9% NaCl. However, the corrosion of Zr was observed to be unaffected by the presence of H2O2 and/or albumin in 0.9% NaCl. Furthermore, TiZr alloys have been shown to possess enhanced passivity in comparison to CP Ti in the various exposure conditions including highly acidic (HCl), oxidative environments (H2O2 in 0.9% NaCl) and cell cultures (macrophage). Corrosion products of Zr were characterised in situ by synchrotron X ray methods, which were found to be ZrOCl2 ∙ 8H2O, tetragonal ZrO2, and Zr metal fragments in 0.9% NaCl regardless of the presence or absence of H2O2 and/or albumin. The presence of Zr metal fragments as a result of a corrosion process indicates the generation of metal species in the absence of wear.

Analysing the voice behind the words represents a central aspect of human communication and is thus the key to intelligent machines such as social robots and virtual assistants. In the realm of Computational Paralinguistics (CP), research has aimed at more natural human-computer interaction by enabling machines to recognise various speaker characteristics, social signals and other non-verbal cues. Shaped by the standard methodology in machine learning, research in CP has devised single-task learning systems for recognising dozens of speaker attributes. However, human-like machine perception demands the holistic analysis of paralinguistic phenomena, learning how these are interrelated and using this knowledge to gain a better understanding of the whole context. Thus, in this thesis, two major challenges in current research are addressed: (i) Overcoming the bottleneck of data annotation to meet the urgency for obtaining labelled data; (ii) Exploiting task relatedness in learning multiple tasks at the same time. To tackle the first problem, efficient data annotation methods based on active learning and semi-supervised learning are proposed, with the aim to leverage the vast amount of unlabelled data for training while reducing manual labelling effort. In determining whether or not and how many human annotators are required in regard to label ambiguity, a generic cooperative learning framework is introduced, using measures of model confidence for human-machine arbitration and an early-stopping criterion based on interrater agreement. Techniques of this kind are particularly important for subjective tasks that do not come with ground truth labels. The effectiveness of the proposed algorithms is demonstrated on the tasks of speech emotion recognition and prosody estimation. To enable holistic analysis in meeting the second thesis objective, it is important to know which related tasks may mutually benefit from joint learning and thus to avoid negative transfer. Building upon feature relevance analysis and non-metric dimensional scaling, a data-driven approach is proposed to construct a paralinguistic atlas, in which task similarities are transformed into distance measures in a two-dimensional space. On this basis, various multi-task learning methods are applied to related tasks to improve generalisation performance by means of data aggregation, feature relevance analysis, auxiliary labels, and shared representation learning. In particular, two approaches have been conceived to handle the prevalence of single-label databases, which represents a main obstacle for holistic analysis. One possibility to exploit task interdependencies is to generate multi-label datasets. This is achieved with the proposed cross-task labelling method, which automatically completes the missing target labels, e. g., when performing joint learning of deception detection and sincerity classification. Alternatively, the single-label datasets at hand can be fed into a multi-task deep neural network, which learns different tasks in parallel through shared hidden layers and separate output layers. This method also effectively addresses the 'universality' aspect of the holistic concept, i. e., recognising various paralinguistic patterns such as different emotion expressions in a simultaneous way.

Due to the advantages of high power density, high efficiency and compact size, high speed permanent magnet machines (HSPMMs) have found wide application in industrial areas. Compared with a conventional speed permanent magnet machine, a HSPMM rotor can reach speeds of more than 10,000 rpm, which brings challenges with regard to electromagnetic, thermal and mechanical aspects of machine design. The higher power density also results in larger power loss per unit volume; due to the small machine size, machine thermal dissipation becomes difficult. Moreover, air frictional loss rises dramatically when the rotor is in high speed operation and this may also further increase rotor temperature. Therefore, research into HSPMM power losses and improving machine thermal dissipation capability is of significant interest. HSPMM mechanical issues also need to be considered to ensure safe and reliable machine operation. As rotor speeds rise, rotor strength becomes prominent and critical as the permanent magnets are vulnerable to the large centrifugal force. In addition, the machine rotor should also have enough rigidity and avoid operating at critical speeds. As such, this dissertation focuses on HSPMM design and research. Multi-physical fields analysis of a HSPMM is carried out to calculate machine power losses and temperature distribution, with factors influencing machine performance considered; HSPMM rotor mechanical research and analysis are also carried out and presented in this study. Firstly, the HSPMM design methodology and process are illustrated with machine rotor parameters, PM material, pole numbers and rotor sleeve considered for a 150 kW, 17000 rpm HSPMM. Then, HSPMM performance for different machine stator structures and PM pole arc pole pitches is investigated using the Finite Element Method (FEM) for the machine operating at both no load and full load conditions; HSPMM electromagnetic performance and how it is impacted by machine parameters is also studied. HSPMM power losses are comprehensively investigated in the following chapter. As machine core loss can be significantly increased with increasing machine frequency, it is critical to accurately estimate HSPMM iron loss. Based on the machine iron core magnetic field variation that is obtained by FEM analysis, machine steel iron core loss estimation for HSPMM is performed using an improved method with the influences of alternating and rotating magnetic fields, as well as harmonics effects, considered for high precision. Then the HSPMM air gap magnetic flux density distribution considering machine stator slotting effect is also analytically calculated with its effectiveness verified by FEM results. Then rotor eddy current loss is studied by time-stepping FEM, while the effects of rotor sleeve dimensions and properties, copper shielding composite rotor structure, air gap length, as well as slot opening width are further researched in depth. A PM bevelling method is also proposed and investigated to reduce HSPMM rotor eddy current loss while having little effect on machine output torque. Then a fluid field analysis is carried out to study HSPMM rotor air frictional loss when the rotor is in high speed operation. According to the characteristics of a machine axial forced air cooling system, the HSPMM temperature distribution is investigated by 3-D fluid–thermal coupling CFD modelling with the calculated power losses results. The machine thermal analysis theory and modelling method are also detailed and further explained. HSPMM thermal performance variation due to impacting factors of cooling air velocity, rotor eddy current loss and sleeve thermal conductivity are also comprehensively investigated and studied in this dissertation. The designed HSPMM is prototyped, and temperature experimental tests are also carried out to verify the effectiveness of the research and analysis for HSPMM. Then, thick-walled cylinder theory is introduced to study rotor mechanical strength analytically, while it also verifies the FEM calculation results. Then based on FEM analysis, HSPMM rotor stress distribution is investigated with sleeve material effects on rotor strength discussed. In order to alleviate the rotor sleeve stress, three pole filler materials are comparatively studied, while the temperature impacts on rotor mechanical stress is further considered; sleeve thickness and the interference between PM and sleeve are investigated in an integrated fashion for HSPMM rotor strength analysis, with some conclusions also drawn for HSPMM rotor mechanical design. HSPMM rotor critical speeds are also calculated by the established 3D rotor dynamic analysis FEM model to ensure the rotor is operating in a desirable condition.

This project investigates the impact and plasticity of perceptual cue weighting strategy for normal hearing (NH) listeners with cochlear implant (CI) acoustic simulation and CI users. It is hypothesised that how listeners allocate perceptual attention on different speech cues is related to how accurately and effectively they can restore the phonemic structures from the acoustic inputs. Therefore, it can be beneficial to use auditory training to guide listeners’ attention to the more reliable and informative cues for their own specific language, in order to improve speech recognition and ease listening effort. The first part of this project investigated the impact of perceptual weighting strategy for both groups of listeners. Listeners’ sentence recognition and pupillary response (taken as a measure of listening effort) were measured. They were then taken into the same model with listeners’ acoustic cue weighting ratio and auditory sensitivity to explore their relation. The second aim of this project was to examine the possibility of using auditory training to change listeners’ acoustic cue weighting strategy towards an optimal one. A distributional training method was used here, with the sampling of spectral and temporal cues in the training word stimuli manipulated in a way that only the spectral cue followed a bimodal distribution that resembled the natural speech. This was to increase the saliency of the spectral cue, in order to direct listeners’ attention to the spectral cue. Sentence recognition performance in quiet, acoustic cue weighting strategy and pupillary responses were measured before and after the training to examine the effectiveness of the training. Findings from this project will extend the current understanding on CI users’ perceptual cue weighting strategy and provide inspiration for a more comprehensive rehabilitation scheme for CI users.

This thesis is composed of two parts: Part I is focused on studying how quantum dot (QD) surface small-molecule capping ligands affect its Förster resonance energy transfer (FRET) with a fluorescent protein (FP); Part II is focused on developing an ultrasensitve DNA sensing technology by combining magnetic nanoparticle (MNP) and enzymatic signal amplification. Part I The FRET between a CdSe/ZnS core/shell QD capped with three different small-molecule ligands and a hexa-histidine (His6)-tagged FP (mCherry) has been studied. Results show that small-molecule ligands strongly affect the FRET behaviours between QD and FP. The QD-FP self-assembly process is fast (complete in minutes at low nM concentration), strong (with Kd ~ 1 nM), suggesting that the QD-His6-tagged biomolecule self-assembly is an effective approach for making compact QD-bioconjugates which may have a wide range of sensing and biomedical applications. PART II I have developed a facile, rapid, and sensitive DNA sensor by combining the efficient MNP-based target capture with the highly efficient signal amplification power of enzymes to achieve ultra-sensitivity. This sensor works efficiently in both pure buffer and complex media (such as 10% human serum in buffer). Moreover, this developed approach is able to quantitate two distinct DNA strands in a homogenous phase at the same time with a detection limit of ~5 pM. Second, I have developed a novel, highly sensitive and selective approach for label-free DNA detection and single-nucleotide polymorphism (SNP) discrimination by combining target-recycled ligation (TRL), MNP assisted target capture/ separation, and efficient enzymatic amplification. This approach possesses a detection limit of 600 fM unlabelled DNA targets and offers exquisitely high discrimination ratio (up to > 380 fold) between a perfect-match mutant and its single-base mismatch DNA target. Furthermore, it can quantitate the rare cancer mutant in large excesses of coexisting wildtype DNA down to 0.75%.

The biosorption of Cu(II), Pb(II), Zn(II) and Ni(II) was investigated both individually and from mixed metal ion solutions in batch and packed-bed flow-through column experiments using the hydrophilic polyurethane immobilised biomass matrix. Initial work tested the biosorption stability of immobilised Sphagnum moss in bath experiments. The polyurethane immobilisation technique helped moss maintain around 90% of its biosorption capacity over 10 repetitive cycles of biosorption/desorption using HC1 as a desorbent with typically more than 95% metal ion recovery from each cycle. Factors which it was thought might affect the performance of biosorption were also explored, including solution pH, other metal ions, organic materials and biomass particle size. Further batch experiments using Sphagnum moss look at sequestration of Cu(II), Pb(II), Zn(II) and Ni(II) both individually and from mixed metal ion solutions. The data were evaluated using the Langmuir isotherm equation, and it was concluded that care must be taken in presentation and interpretation of results when this modelling approach is applied at low concentrations. Repetitive metal loading cycles also gave lower values for biosorption capacity compared to the maximum potential value, due to equilibrium effects. Biosorption from a multi-component metal system showed that lead and copper competed equally for binding sites and much more effectively than zinc and nickel. Equations ere derived to predict the percentage effect on a given metal ion of other metals in a multi-metal system based on their behaviour in the relevant single and binary systems. Biosorption column experiments were carried out on multicomponent mixtures of Cu(II), Pb(II), Zn(II) and Ni(II) using immobilised Sphagnum moss, seaweed Ascophyllum nodosum, sunflower waste, and whole plant maize. Columns with seaweed held 0.68 mmol g⁻¹ metal, followed by Sphagnum moss (0.30 mmol g⁻¹) and sunflower waste (0.26 mmol g⁻¹). Binding by maize was poor at 0.02 mmol g⁻¹ metal and it was considered unsuitable as a biosorbent. Titration experiments revealed differences in the number of acidic functional groups on the different types of biomass and a recommendation was made that a simple titration experiment should be the first step for biosorbent screening.

Multidomain environments where multiple organizations interoperate with each other are becoming a reality as can be seen in emerging Internet-based enterprise applications. Access control to ensure secure interoperation in such an environment is a crucial challenge. A multidomain environment can be categorized as "tightly-coupled" and "loosely-coupled." The access control challenges in the "loosely-coupled" environment have not been studied adequately in the literature. In a "loosely-coupled" environment, different domains do not know each other before they interoperate. Therefore, traditional approaches based on users' identities cannot be applied directly. Motivated by this, researchers have developed several "attribute-based" authorization approaches to dynamically build trust between previously unknown domains. However, these approaches all focus on building trust between individual requesting users and the resource providing domain. We demonstrate that such approaches are inefficient when the requests are issued by a set of users assigned to a functional role in the organization. Moreover, preserving "principle of security" has long been recognized as a challenging problem when facilitating interoperations. Existing research work has mainly focused on solving this problem only in a "tightly-coupled" environment where a global policy is used to preserve the "principle of security." In this thesis, we propose a role-based access control and trust management framework for "loosely-coupled" environments. In particular, we allow the users to specify the interoperation requests in terms of requested permissions and propose several "role mapping" algorithms to map the requested permissions into roles in the resource providing domain. Then, we propose a "Simplify" algorithm to simplify the distributed proof procedures when a set of requests are issued according to the functions of some roles in the requesting domain. Our experiments show that our "Simplify" algorithm significantly simplifies such procedures when the total number of credentials in the environment is sufficiently large, which is quite common in practical applications. Finally, we propose a novel "policy integration" approach using the special semantics of "hybrid role hierarchy" to preserve the "principle of security." At the end of this dissertation a brief discussion of implemented prototype of our framework is present. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

There is a need to address disparities in the underrepresented minority (URM) STEM students’ degree attainment and retention, vs. non-URM students. A literature review suggests that URM STEM students face barriers related to demographic, academic, and social-cognitive factors. The Louis Stokes North Star STEM Alliance (LS-NSSA) seeks to address these factors and promote URM student success in STEM disciplines. However, there is a gap in knowledge regarding LS-NSSA’s mechanism and outcomes. The current study investigates factors influencing URM STEM students’ retention and graduation, investigates LS-NSSA program effectiveness, and investigates the path by which outcomes are achieved. A theory-based evaluation approach is utilized in an explanatory sequential mixed-methods design guided by LS-NSSA’s theory of change. The results of three sub-studies suggest the importance of first-semester experience and academic outcomes for URM STEM students’ graduation and retention. Participation in LS-NSSA is associated with a higher level of academic preparation to pursue a bachelor’s degree, compared to URM students not affiliated with the program. URM STEM students participating in LS-NSSA’s research mentorship programs have higher levels of confidence, interest, science identity, sense of belonging, and commitment to STEM. The current study offers evidence supporting LS-NSSA’s theory of change regarding student academic and social-cognitive trajectories. The current study suggests directions for future evaluative studies of LS-NSSA.

Direct current power distribution and dc microgrids have been gaining momentum in recent years for various applications. However, compared to traditional ac systems, dc systems still need to address technical challenges such as fault current management/protection, power flow control, power quality management, and potential system instability. This dissertation proposes the T-Breaker system seeking to address these issues in an all-in-one device with modular multilevel converter functions. It is characterized by its use of locally integrated energy storage, high scalability, and high tolerance to control signal mismatch during the fast network transients. This is a paradigm shift from traditional solid-state circuit breakers as the proposed T-Breaker not only protects against faults, but can also function as an energy router with unparalleled ancillary functions for dc grids.This dissertation starts with a systematic overview of the remaining challenges of dc distribution covering the existing dc circuit breaker technologies, particularly on solid-state circuit breakers, as well as the needs and current status of compensation in dc networks.Taking inspiration from the series and shunt compensation devices in ac transmission and distribution systems, the T-Breaker system with half-bridge sub-modules is derived from traditional solid-state circuit breakers. The proposed circuit has modular multilevel converter functions with an increased number of active switches but no conduction loss penalty when compared with traditional solutions. The basic operation modes and the circuit analysis are carried out, and the limitations are identified for the half-bridge T-Breaker. In order to improve upon these drawbacks, the full-bridge T-Breaker is proposed, with improved performance in sub-module voltage injection to the line, and lowering of total heat flux for more compact thermal designs. Along with the basic functionalities of both T-Breaker topologies, the series and shunt compensation functions are also introduced and demonstrated with simulation results.Extended discussions are carried out for the two active current limiting methods in T-Breaker: modulated voltage injection and driving the semiconductor devices into current clamping regions. The discussion includes equivalent circuit models, circuit simulations, and test results. A review of existing solutions and literature of fault current limiters are also provided.Two prototypes were developed at 1 kV, 500 A and 20 kV, 250 A in order to demonstrate the functionalities of the T-Breaker for different applications and voltage levels. Targeting the high-current aircraft and commercial building applications, the 1-kV, 500-A, half-bridge T-Breaker system is demonstrated with 99.58% conduction efficiency at full load and up to 5 kA of current breaking capability. Targeting electric ships and utility scale dc microgrids, the full-bridge T-Breaker is demonstrated at 20 kV and up to 250 A of conduction current. The detailed component sizing and design procedures are discussed. The key design challenges, including paralleling power modules for 500-A operation, assembly-level insulation design with 3D printed mechanical housings for medium voltage, and high current, medium voltage printed circuit board bus bar designs with specialized solder masks and terminal arrangement, are also presented.

The physicochemical properties of the secondary organic materials (SOMs) that constitute the particle phase have potentially important consequences for the growth, the reactivity, and ultimate fate of atmospheric organic aerosols, thereby affect climate, human health and visibility. A quantitative analysis of the physicochemical properties of the SOMs is important, but challenging. This thesis presents laboratory studies of α-pinene derived SOMs, which is one of the major components of secondary organic aerosols (SOAs) in the forests, by combing a flow tube reactor, aerosol particle mass analyzer (APM) and other online/offline measurement techniques. A water-jacketed constant temperature flow tube reactor was built to produce SOM particles grown from either condensation or coagulation. Different ratios of α-pinene enantiomers were mixed and injected into the flow tube reactor for dark ozonolysis. A matrix of organic precursor and ozone concentrations was designed and tested in order to determine the optimal concentration to switch between condensation and coagulation. Results show that at 51 ± 1 ppm O3, condensation is the dominant growth mechanism when the α-pinene concentration is 0.125 ± 0.001 ppm, and coagulation is the dominant growth mechanism when the α-pinene concentration increases to 1.00 ± 0.03 ppm. A combination of both growth mechanisms is observed when the α-pinene concentration is in between the described values. The study also proposed and tested the hypothesis that a 50:50 mixture of α-pinene enantiomers may result in SOM particles that have different physical properties, such as number-diameter distributions, when compared with those particles generated from a single enantiomer of α-pinene. The experiment was conducted within the condensational growth regime so that the chirality induced structure differences in oligomers can be maximized during the nucleation and condensation. Nevertheless, our analysis indicates that, after removing the effects of ozone and temperature, the chirality-induced effects are minimal and within our detection limit. Even though the results were negative, the method used in this experiment provided useful experience for the viscosity related experiments in this thesis. Another important property of the SOM is its viscosity. The viscosities of atmospheric particles determine whether their interactions with surrounding gases are confined to the surface or can proceed to the interior. Viscosities affect the gas-particle diffusion rate, and ultimately influences the SOM’s other physical properties, such as particle size, and chemical properties, such as reactivity. The work presented in this thesis estimates the viscosity of submicron organic particles while they are still suspended as an aerosol without further post-processing techniques that can have the possibility of altering the properties of semivolatile materials. The results show that the studied particles are semisolid up to 58% relative humidity (RH) and may become liquid only at a higher RH. These results imply that atmospheric particles, at least those similar to the ones studied and for low to middle RH regimes, are expected to reach equilibrium only rather slowly with the chemical composition of the gas phase, sometimes on timescales longer than the actual residence time of the particles in the atmosphere. Last but not the least, the results of offline particle analysis from two collaboration studies are also discussed in this thesis. The results show how water vapor, or RH, affects the physicochemical properties of the α-pinene derived SOM particles. In one study, the diffusivity is underestimated by approximately 8 orders of magnitude if calculated from the Stokes-Einstein equation, which suggests the breakdown of Stokes-Einstein equation for small gas molecules. The second study shows the surface properties of SOM particles can be influenced by the RH, leading to a difference of the gas-particle interactions at the particle surface., Engineering and Applied Sciences - Engineering Sciences

Computer graphics arts such as animations, video games, and special effects in live-action movies have become essential for people seeking entertainment and education. This study aims to explore the potential for experimental film in presenting scientific theory as well as assessing different production strategies in 3D image creation. To invite people into some abstract or complicated scientific topics more readily, non-narrative film form is a viable method to relay this type of information. It's crucial to look at how independent filmmakers employ various ways to fulfill their particular creative purposes. I’ll be demonstrating how these processes worked in making my film, Discontinuity, a short 3-D animated experimental work that attempts to illuminate some of the mysteries of quantum theory for an audience. I plan to use my analysis of the film’s production time, the overall quality and the feedback it received to build ideas for future research as well as an overall vision for computer graphics arts.

Computational chemistry has a variety of applications, from understanding a phenomenon of an experiment, to broader applications such as material science, drug discovery, biocatalysis, and many other fields. The following chapters are focused on using computational chemistry tools to understand the mechanisms of enzymes or organic reactions, and to engineer enzymes that produce novel products. Different methods have been applied to achieve these goals, such as molecular mechanics methods, ab initio molecular dynamics, classical molecular dynamics, quantum mechanics (density functional methods), monte carlo methods and machine learning methods. The first chapter is focused on computational modeling of terpene synthase family, to understand the catalysis process of terpene synthase, and further engineer terpene synthase to produce unnatural products. The second chapter is focused on studying organic reaction mechanisms and catalysis using quantum mechanics and molecular dynamics methods. The third chapter is focused on protein structure prediction of terpene synthase.Chapter 1. Mechanistic studies and enzyme engineering of terpene synthase Terpene or terpenoid is the largest class of natural product, with more than 80 000 members. All of which derives from the simple five-carbon isoprene unit. The five-carbon unit can be ligated by prenyltransferases to produce C 10 , C 15 , C 20 … linear structures. These linear structures can then be cyclized by terpene synthase (TPS) to produce multi-ring, multi-stereocenter, cyclic structures. This process usually involves multiple highly reactive carbocation intermediates, brings challenges for modeling and identifying productive binding orientation. Terdockin method is applied to study the catalysis mechanism of the diterpene synthase Rv3377c, as well as ent-kaurene synthase BjKS. Results reveal the mechanism for catalysis, which can be further applied to engineer terpene synthase to produce unnatural products. A few engineer efforts listed in the chapter was made to alter the product outcome for BjKS.Chapter 2. Quantum mechanics studies of organic reactions In the first section of this chapter, the source of the rate acceleration for carbocation cyclization in a biomimetic supramolecular cage is studied using quantum mechanics methods, molecular dynamics, and QM/MM methods. Previous experimental results indicate that the supramolecular cage increases the nazarov reaction rate by roughly 1 000 000 times. The electrocyclization step is slightly enhanced by the catalyst, suggested by computational modeling. The major contribution, indicated by the QM/MM studies, is the activation of the leaving groups, similar as terpene synthase activate the diphosphate leaving group.Chapter 3. Protein structure prediction of terpene synthase Terpene cyclases catalyze one of the most complex chemical reactions in biology, converting simple acyclic oligo-isoprenyl diphosphate substrate to complex polycyclic products via carbocation intermediates. Many computational studies were carried out to illuminate the structural-function relationship of terpene cyclases, which generally rely on a crystal structure. However, among 15,000 terpene cyclases sequences, there are only about 30 types of terpene cyclase crystal structures. To fill in this gap, we proposed a comparative modeling approach to generate high resolution models of class I terpene cyclase, which can be further used as a starting point to predict the productive binding mode of the substrate and potentially set stage for the rational engineering of terpene cyclases.

This dissertation studies the role of peer disclosures in shaping corporate R&D investments. Using the online registration of clinical trials in the drug development process, I find that a firm’s R&D investments are deterred by disclosures of clinical trial initiation from strong rivals but encouraged by disclosures from weak rivals. The cross-sectional analyses suggest that the deterrence effect of peer disclosure is stronger when the therapeutic area has a high clinical-trial success rate, the encouragement effect is stronger when the market has fewer competing firms, and both effects are strengthened when the focal firm has a diversified R&D portfolio. Overall, my findings suggest that the way a firm reacts to peer disclosure varies with the disclosing firms’ relative competitiveness in the R&D race.

We construct an invariant called guts for second homology classes in irreducible 3-manifolds with toral boundary and non-degenerate Thurston norm. We prove that guts of second homology classes in each Thurston cone are invariant under a natural condition. We show that the guts of different homology classes are related by sutured decompositions. As an application, an invariant of knot complements is given and is computed in a few interesting cases.The minimal volume of orientable hyperbolic manifold with a given number of cusps has been found for 0,1,2,4 cusps, while the minimal volume of 3-cusped orientable hyperbolic manifolds remains unknown. By using guts in sutured manifolds and pared manifolds, we are able to show that for an orientable hyperbolic 3-manifold with 3 cusps such that every second homology class is libroid, its volume is at least 5.49... = 6 times Catalan’s constant.In the final chapter, we develop a method for controlling the upper bound of the Euler characteristic of surfaces in sufficiently long Dehn fillings. By using this method, we show that for a compact orientable irreducible acoannular 3-manifold with toral boundary, properly norm-minimizing surfaces capped off with disks in sufficiently long fillings are still properly norm-minimizing. We also show that if a sufficiently long Dehn surgery on a link in a product sutured manifold yields the same product sutured manifold, then this link is horizontal in the product sutured manifold. Combining these results, we prove that for a compact orientable irreducible acoannular 3-manifold with toral boundary, the cores of its sufficiently long fibered Dehn fillings are either horizontal or vertical. Moreover, we provide a new way to prove that there are finitely many links with a given complement such that no component is unknotted and no 2-component sublink is coaxial. Another result coming from this method is that all sufficiently long fillings of an orientable, irreducible, boundary-irreducible-irreducible, acoannular 3-manifold with toral boundary are still irreducible, boundary-irreducible-irreducible, acoannular.

Intelligent transportation systems involve a variety of information and control systems methodologies, from cooperative systems which aim at traffic flow optimization by means of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, to information fusion from multiple traffic sensing modalities. This thesis aims to address three problems in intelligent transportation systems, one in optimal control of connected automated vehicles, one in discrete-event and hybrid traffic simulation model, and one in sensing and classifying roadway obstacles in smart cities. The first set of problems addressed relates to optimally controlling connected automated vehicles (CAVs) crossing an urban intersection without any explicit traffic signaling. A decentralized optimal control framework is established whereby, under proper coordination among CAVs, each CAV can jointly minimize its energy consumption and travel time subject to hard safety constraints. A closed-form analytical solution is derived while taking speed, control, and safety constraints into consideration. The analytical solution of each such problem, when it exists, yields the optimal CAV acceleration/deceleration. The framework is capable of accommodating for turns and ensures the absence of collisions. In the meantime, a measurement of passenger comfort is taken into account while the vehicles make turns. In addition to the first-in-first-out (FIFO) ordering structure, the concept of dynamic resequencing is introduced which aims at further increasing the traffic throughput. This thesis also studies the impact of CAVs and shows the benefit that can be achieved by incorporating CAVs to conventional traffic. To validate the effectiveness of the proposed solution, a discrete-event and hybrid simulation framework based on SimEvents is proposed, which facilitates safety and performance evaluation of an intelligent transportation system. The traffic simulation model enables traffic study at the microscopic level, including new control algorithms for CAVs under different traffic scenarios, the event-driven aspects of transportation systems, and the effects of communication delays. The framework spans multiple toolboxes including MATLAB, Simulink, and SimEvents. In another direction, an unsupervised anomaly detection system is developed based on data collected through the Street Bump smartphone application. The system, which is built based on signal processing techniques and the concept of information entropy, is capable of generating a prioritized list of roadway obstacles, such that the higher-ranked entries are most likely to be actionable bumps (e.g., potholes) requiring immediate attention, while those lower-ranked are most likely to be nonactionable bumps(e.g., flat castings, cobblestone streets, speed bumps) for which no immediate action is needed. This system enables the City to efficiently prioritize repairs. Results on an actual data set provided by the City of Boston illustrate the feasibility and effectiveness of the system in practice.

Designing a biomimetic nanoparticle is a challenging topic. Despite the advances in surface chemistry and material science, it’s still impossible to fully replicate the complex interface that presents in nature. Besides, the bottom-up method fails to prevent the exposure of foreign material when administrated in vivo. To solve this challenge, we directly fuse the natural cell membrane onto the synthetic core. This top-down method quickly generates natural stealth for the synthetic material and preserves the biofunction from the resource cell. Due to its unique properties, cell membrane coating technology displays great potential in the field of drug delivery, detoxification, vaccination and biosensing. In my dissertation, I work on engineering multi-dimensional cell-membrane coated nanoparticles (also known as “nanosponge”) to treat bacterial infection. Instead of direct inhibiting the bacteria growth, nanosponge targets the virulence factors that bacteria secrete to subvert immune surveillance, which is known to place less evolutionary pressure to induce the antibiotic resistance. In the first thrust, antivirulence efficacy is carried out solely by biomimetic nanosponges. Two types of cell membrane from different origins are explored in this direction. In the second thrust, nanosponges are mixed with the supporting material to form a stable composite, which in turn boosts the antivirulence efficacy of the nanosponge. Together, the cell membrane coating technology offers us a great tool to solve the clinical challenges by mimicking the biological events in the natural biointerface.

Microbial electrolysis system (MES) is a promising technology which integrates microorganisms or other bio-catalysts with an electrochemical system to generate electricity or chemicals including energy carriers such as methane. Compared with the combined anaerobic digestion and CO₂ biomethanisation system, MES in principle requires lower energy input for methane production especially when coupled with organic material oxidation. This study investigated the simultaneous CH₄ production and organic substance degradation in MES, using synthetic swine wastewater as the model substrate. The results demonstrated that the applied cathodic potential played a crucial role for CH₄ production in the biocathode and influenced to a great extent the reactions and operational conditions in the bioanode. The performance of the anode, therefore, could not be controlled exclusively by the organic substance loading applied to this compartment. Abiotic anodes were then evaluated for the organic degradation to alleviate the impact of acidification on anode performance. Although the boron doped diamond (BDD) showed higher rate of organic substance degradation compared to platinum due to its high oxygen overpotential, O₂ evolution was still observed as a side reaction during the glucose oxidation process. A middle chamber for O₂ removal using N₂ stripping was then introduced to control the O₂ transfer to the biocathode: the limited reduction in cathodic O₂ concentration was achieved at the expense of the complex middle chamber operation and the increased internal resistance. The three-chamber MES therefore resulted in a lower volumetric CH₄ production, when compared to the dual-chamber cell. The high pH of the cathodic chamber in both dual-chamber and three-chamber cells indicated an opportunity for plant nutrient recovery, based on a theoretical model simulation.

Solid waste represents a threat to the environment and human health. Organic residues can be converted to valuable products through biological processes, reducing the volume and risks associated to the degradability of the waste. Acidogenic fermentation (AF) is the mixed-culture microbial conversion of organics to volatile fatty acids (VFA), such as acetic, propionic, and butyric acids. VFA are chemical platforms used for the synthesis of high value-added products. There is currently a lack of understanding on mechanisms to control pathways to avoid unwanted by products; and maximise VFA yields. In this thesis, organic residual solids (ORS) have been used as substrate to study the effect of different fermentation conditions such as pH, hydraulic retention time, carbon to nitrogen ratio, CaCO3 addition, urea addition, feeding mode, inoculum pre-treatment, and enzymatic pre-treatment. It was found that fermentation conditions highly influence the type of VFA generated and the VFA yield. Fed-batch feeding mode resulted in increased VFA concentrations and yields; and longer chain fatty acids were produced. A large-scale process was also proposed followed by an economic assessment of the AF process using ORS as substrate, further demonstrating the economic potential of biological waste conversion to high-value products. The assessment concluded that most of the variable operational cost is due to feedstock pre-treatment and product recovery, due to the high cost of enzymes and the energy required for VFA recovery.

A rate-based PZ aerosol growth model was developed in gPROMS [superscript ®] ModelBuilder. Amine Aerosol growth was simulated at the unique conditions of piperazine (PZ) and the pilot plant absorber configurations at the National Carbon Capture Center. Amine aerosol growth is driven by amine-limited diffusion. As aerosol concentration increases, aerosol growth decreases due to the depletion of the amine driving force in the gas phase. Aerosol growth can be increased by enhancing the gas-film mass transfer coefficient of packing. A solvent with moderate volatility, like PZ, will produce aerosol that grows to larger size and is easier to collect. Solvents with low volatility should be avoided as they produce aerosol that is hard to collect. Process configurations that provide greater water partial pressure in the water wash, such as higher operating temperature and pre-humidified empty space, will help aerosol grow. Two pilot plant campaigns were designed and conducted in this work. 5 molal (m) PZ was operated for the first time and provided significant absorber performance benefits over 8 m PZ due to enhanced mass transfer rates from lower solvent viscosity. Parametric tests were performed with a wide range of absorber operating conditions. With the existing model correction, the pilot plant absorber model could reasonably capture the measured absorber performance. For future campaigns, this work recommended that the pilot plant absorber should be operated at both pinched and not pinched conditions. Both equilibrium correction (correct for errors in solvent loading measurements and effects of degradation) and packing correction (correct for effects of rivulets and drop and additional mass transfer caused by distributors and chimney trays) should be utilized in the data reconciliation process. A membrane-amine hybrid carbon capture system for natural gas combined cycle (NGCC) power plants was proposed and evaluated. When the inlet CO2 increases from 4% to 18%, the total absorption costs decrease by 60% and the total regeneration costs remain the same. Amine scrubbing without the direct contact cooler was found to be a superior design for NGCC carbon capture. The absorber gas inlet must be designed to avoid excessive localized temperature and solvent evaporation.

Visual art is probably the oldest art. But people with vision challenges have been held back, perhaps for many years by avoiding making art. However, it is never late to start and this painting system helps to paint with ease and joy. Through the researching about the behavior of people with visual disability, the conclusions are: Painting can help people with limited vision to face life in a more positive way; vision and touch are closely related. After the shape symbolism experiment and interview, the conclusion is people with visual disability understand the metaphors of the pictures. The mock blind life experience shows that people with visual disability have heightened other senses. After the attempts of blind drawing, the problems came out and the design goals are help people with visual disability draw accurate position and size, recognize colors and set up the supplies independent. The outcome is a painting system that allows people who are blind or who have limited vision to draw, while also allowing people with sight to see what is being drawn. The user simply presses the raised guideline and uses tactile guide points to draw. The easel also features a folding, latching storage drawers and a textured palette with for mixing pigments, a sink, and cleaning foam. The brushes with convenient textured tags in their handles, allow one to identify the shape of brushes. The six basic color pigments are scented, which helps to recognize them by smell, on top of braille labels. The raised guide arrows can be touched everywhere in the box for tactile reference and guidance.

Natural images are usually either sparse by themselves or sparse after a transform. For instance, images of stars from telescope are sparse by themselves and images of humans are sparse under the wavelet transform. Sparsity plays an important role in image processing and machine learning. How to build appropriate sparse models, how to numerically find solutions from the sparse models and how to derive theoretical guarantees of the correctness of the solutions are essential for the success of sparsity in applications. In this thesis, we study sparse modeling in image processing and machine learning. We propose a novel two-stage (projection followed by correction) modeling framework for image reconstruction from nonuniform Fourier measurements. We design domain adaptive sparsity regularization schemes that are able to maintain image smoothness as well as edge details. Various numerical results in both one-dimensional signals and two-dimensional images are also presented to demonstrate the superior performance of the proposed model compared with other methods.We study the problem of learning data adaptive overcomplete bases so that each data example can be linearly represented by few basis vectors. In the specific case when the input data size is large, we design a novel distributed dictionary learning algorithm which builds on the classical consensus approach. Most importantly, the proposed algorithm imposes a structured constraint to ensure that the derived dictionary has low coherence, that is, the columns of the dictionaries are nearly orthogonal. We analyze the solution obtained from the well known Least Absolute Shrinkage and Selection Operator (Lasso) model in the feature selection scenario. A substantial challenge in feature selection to specify the relation between the factors/features and the observations (e.g., linear or nonlinear). The relation is however often unknown. We prove that under certain conditions, Lasso is able to find the correct features (asymptotically) even though the model is misspecified. The contributions of this thesis are summarized below: 1. We propose a new two-stage model with domain adaptive sparsity constraints for image reconstruction from nonuniform Fourier measurements.2. We design a new distributed dictionary learning algorithm.3. We prove a new theorem on the correctness of Lasso model in sparse feature selection when the relationship between observations and features is misspecified.

Nowadays, the addition of nanoparticles (NPs) in polymer has attracted intensive attention because nanoparticles can bring some excellent properties to polymer materials. To get better control of the dispersion of NPs, polymer-grafted nanoparticle (PGNP) has been used in this work because the polymer ligands on the surface of NPs can give phase separation in the system. The phase separation behavior of bianary PGNP blend thin films has been investigated in this work. The blend thin film is composed of PS-g-SiO2 and PMMA-g-SiO2 nanoparticles. The phase-separated domain growth was slower than PGNP blends with shorter grafted chain lengths. With the application of capillary force lithography (CFL), more PMMA-g-SiO2 nanoparticles were segregated in imprinted trenches with longer thermal annealing time. In contrast, faster soft-shear cold-zone-annealing (CZA-SS) speed induced selective segregation of PMMA-g-SiO2 particles. The process under CFL is a wetting-driven process and that under CZA-SS is a shear-driven process.

Acute Myloid Leukemia (AML) is one of the most common leukemia in adults with poor five years survival even with modern chemotherapy treatment. New strategies need to be developed in order to enhance the survival of AML patients. Interferon-α (IFN-α), a type I IFN, is a known as an anti-tumor agent, which used to clinical treat AML because of its anti-proliferative effect and immune response. However, the clinical trials show divergent results because of the instability and delivery difficulty. USP18 is a negative regulator of Type I IFN signaling pathway. Knockout Usp18 can enhance and prolong the Type I IFN pathway. Here, we use AE9a-included leukemia mouse models that reflect the genetics and pathology of human acute myeloid leukemia to study the effect of Usp18 loss in leukemia development and to discover possible new therapeutic targets for leukemia. Results show that deletion of Usp18 delayed leukemia propagation in both the AE9a-induced and MLL-AF9-induced leukemia models. Usp18-deletion-mediated inhibition of myeloid leukemia propagation is IFN response dependent. More important, Usp18 deletion can delay the myeloid leukemia remission after the chemotherapy. These data indicate that pharmacological inhibition of Usp18 may be benefit in AML treatment with interferon prescribed.

Numerous solar and astrophysical observations of jet- and bubble-like plasma structures exhibit morphological similarities, suggesting that there may be common plasma physics at work in the formation and evolution processes of these structures at different system scales. The ideal magnetohydrodynamics (MHD) provide the necessary theoretical basis for employing laboratory experiments to investigate key physical processes in nonlinear astrophysical and solar systems, especially when magnetic fields are present. A coaxial magnetized plasma gun has been designed, installed, and operated in the HelCat linear device at the University of New Mexico. In Region I, a current-driven plasma jet is formed. The plasma column experiences the current-driven kink instability consistent with the Kruskal-Shafranov criterion. When a perpendicular background magnetic field is applied, the jet column length increases with a longer lifetime and appears to show greater stability. Evidence suggests that magnetic tension, caused by the curvature of background magnetic field, leads to an axial sheared flow, which contributes to the stabilization. The calculated results are accordant with the stabilization criterion - reported by other research groups for similar magnetic geometrics. In Region II, spheromak-like plasma formation is verified. When the spheromak plasma propagates into the background magnetic field, the typical self-closed magnetic configuration does not hold anymore. At the upper-side, the Rayleigh -Taylor instability has been observed. The theoretical analysis is present and the instability growth rate has been calculated.

One of the major problems in treatment of opioid addiction is the repeated reuse and long-term memory of the drug-experience even after prolonged periods of abstinence. During the past decades, there has been enormous expansion in our understanding of how opioid drugs act on the nervous system. A complex brain network including the mesolimbic dopamine system, ventral striatum, extended amygdala, prefrontal cortex and hippocampus is suggested to be associated with the addiction cycle, in particular, the adult neurogenesis taken place in the dentate gyrus (DG) of the hippocampus has a functional implication in opioid addiction. It is intriguing to study the convergence between the modulation of adult neurogenesis and opioid addiction, since the adult-born granule cells were shown to play a role in neuroplasticity of hippocampus function and in the development and retention of drug-contextual memory. In the first part of my study, I attempted to define the temporal window of morphine’s inhibitory effect on adult neurogenesis with a transgenic mouse model. Four days of conditioned place preference (CPP) training with morphine significantly reduced the number of late stage progenitors and immature neurons in the sub-granular zone (SGZ) of mouse hippocampus but did not affect the number of early progenitor cells. The results from colocalization of cell-type selective markers suggested that under the condition of CPP training, morphine affects the transition of neural progenitor/stem cells differentiate into immature neurons. When the transcription factor neural differentiation1 (NeuroD1) was over-expressed in DG by stereotaxic injection of lentivirus, it rescued the loss of immature neurons and prolonged the extinction of morphine-trained CPP. Next, a synthetic small molecule KHS101 which was reported to increase NeuroD1 mRNA in cultured neural progenitors and induce neuronal differentiation in the DG of hippocampus, was utilized to mimic the effect of lentivirus-mediated NeuroD1 overexpression on morphine-primed CPP. The results indicated that subcutaneous injection of KHS101 before conditional training could enhance the retention of drug-related memory and prolong CPP extinction; while the same treatment after conditional training disrupted the drug-contextual associations and shortened CPP extinction. Such KHS101’s effect paralleled that observed when the over-expression of NeuroD1 was temporally controlled with an inducible tetracycline system. Furthermore, the KHS101’s effect could be abolished by the stereotaxic injected NeuroD1 shRNA lentivirus. These studies suggest that morphine decrease the total numbers of newborn neurons in the SGZ by interfering with neural progenitors’ differentiation via a mechanism involving NeuroD1. Since adult neurogenesis serves as an important form of neural plasticity, we assume that certain immature neurons contribute to the formation and consolidation of drug-contextual association memory, and NeuroD1 plays a key role during this process. Such assumption is supported by the observation that compounds such as KHS101 that could regulate NeuroD1 expression in the hippocampus possess the ability to manipulate the extinction of drug contextual memory. In conclusion, the regulation of NeuroD1 activity leads to modulation of adult neurogenesis, thus affecting the drug-association memory.

Interval-censored data arise when failure times cannot be observed exactly but can only be determined to lie within an interval. In this dissertation, our research interest focuses on correlated survival data occur when individuals under study are clustered or experience multiple events of interest. we developed three novel Bayesian Cox models to handle different types of correlated interval-censored survival data. For clustered data, we utilized a shared frailty factor for unobserved correlation between observations within the same cluster. For spatially correlated data, we first used frailty for within cluster correlation, and then assigned a conditional autoregressive distribution prior for frailties to model the spatial dependency between clusters. In the aforementioned two frailty models, we also considered time-varying coefficient for temporal covariate effect. For bivariate data, we applied copula model to account for the dependence between outcomes. Simulation studies and analysis of real data examples illustrate the performance and applications of the proposed methods.

The MapReduce framework is a programming model proposed by Google to process large datasets. It is an efficient framework that can be used in many areas, such as social network, scientific research, electronic business, etc. Hence, more and more MapReduce frameworks are implemented on different platforms, including Phoenix (based on multicore CPU), MapCG (based on GPU), and StreamMR (based on GPU). However, these MapReduce frameworks have limitations, and they cannot handle the collision problem in the map phase, and the unbalanced workload problems in the reduce phase. To improve the performance of the MapReduce framework on GPGPUs, in this thesis, a workload balance MapReduce framework (B_MapCG) on GPUs is proposed and developed based on the MapCG framework, to reduce the number of collisions while inserting key-value pairs in the map phase, and to handle the unbalanced workload problems in the reduce phase. The proposed B_MapCG framework is evaluated on the Tesla K40 GPU with four benchmarks and eight different datasets. The experimental results showed that the B_MapCG framework achieved big performance improvements for all the four test benchmarks both in the map phase and the reduce phase compared with MapCG.

Multituberculates are an extinct mammalian group that has its geological time span from at least the Late Jurassic to the latest Eocene, with fossils on all continents except Antarctica. Before the K-Pg boundary, in most North American Late Cretaceous mammalian faunas, multituberculates outnumbered eutherians and metatherians in both abundance and species richness. After the K-Pg boundary, the taxonomic diversity of multituberculates was succeeded by placental mammals. Multituberculates have unique dentition morphology differing from all extant mammals. Despite the importance of multituberculates in the study of mammalian evolution, much less is known about this enigmatic mammalian lineage compared to our knowledge of therian mammals. My dissertation mainly discusses the phylogenetics of Neoplagiaulacidae, the taxonomically richest family of multituberculates, and reconstructs the diets of cimolodontans (multituberculates from the Late Cretaceous to Paleogene) based on analyses of tooth wear, cusp charpness, and tooth size proportions.My phylogenetic study of neoplagiaulacids use a novel method proposed by me, the standardized character weighting approach. In this method, the variable range of character steps in each different character is set to the same scale. This new method has great superiority compared to the traditional parsimony method, in eliminating character state delimitation problem, finding the most balanced phylogenetic signals among conflicts, and not losing any evidence. The phylogenetic analysis of neoplagiaulacids reveals the evolution pattern of neoplagiaulacids on both large and small scales. The result shows that different lineages had different pathways of morphological specialization and a tendency of convergent evolution in acquiring larger ultimate lower premolar (p4). The result also indicates that ptilodontids were most likely derived from Parectypodus armstrongi. For the diet reconstruction of cimolodontans using multiple lines of evidence, the tooth wear patterns of the ultimate premolars (p4 and P4) are analyzed to infer the tooth function and chewing mechanism, the cusp sharpness on molars are linked with the physical perporty of food items, and the tooth size proportions are used to infer the developmental mode, which further reveals diet signal. The result indicates that the method presented here using combined evidence is robust in inferring the diet of cimolodontans. In the end, the tooth size proportions of multituberculates are compared to that of placental mammals, demonstrating that the lesser radiation and subsequent declination of multituberculates following the K-Pg boundary is correlated with less developmental plasticity of multituberculate teeth compared to that of placental mammals.

The coupling of magnetic and electric order parameters, i.e., the magnetoelectric effect, has been widely studied for its intriguing physical principles and potentially broad industrial applications. The important interactions between ferroic orderings -- ferromagnetism, ferroelectricity and ferroelasticity -- will enable the manipulation of one order through the other in miniaturized materials, and in so doing stimulate emerging technologies such as spintronics, magnetic sensors, quantum electromagnets and information storage. By growing ferromagnetic-ferroelectric heterostructures that are able to magneto-electrically couple via interface elastic strain, the various challenges associated with the lack of single-phase multiferroic materials can be overcome and the magnetoelectric (ME) coupling effect can be substantially enhanced. Compared with magnetic field-controlled electric phenomena (i.e., the direct magnetoelectric coupling effect), the converse magnetoelectric effect (CME), whereby an electric field manipulates magnetization, is more exciting due to easier implementation and handling of electric fields or voltages. CME also affords the possibility of fabricating highly-efficient electric-write/magnetic-read memories. This study involved two avenues of inquiry: (a) exploring the strain-mediated electric field manipulation of magnetization in ferroelectric-ferromagnetic heterostructures, and (b) investigating coupling and switching behaviors at the nanoscale. Accordingly, a series of magnetoelectric heterostructures were prepared and characterized, and their electric field tunability of magnetic properties was explored by various techniques and custom-designed experiments. Firstly, the relevant properties of the individual components in the heterostructures were systematically investigated, including the piezoelectricity and ferroelectric/ferroelastic phase transformations of the ferroelectric substrates, lead magnesium niobate-lead titanate, or Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT). This investigation revealed significant information on the structure-property relationships in crystals oriented at , as well as shed light on the effect of ferroelectric phase transformation on magnetoelectric coupling. This investigation of electric field controlled strain, in contrast to many prior studies, enables a more rational and detailed understanding of the magnetoelectric effect in complex ferroelectric-ferromagnetic heterostructures. The magnetoelectric thin film heterostructures were fabricated by depositing ferromagnetic iron-gallium (Fe-Ga) or cobalt ferrite (CoFe2o4 or CFO) films on top of differently-oriented ferroelectric PMN-PT substrates. Through significant electric field-induced strain in the piezoelectric substrate, the magnetic remanence and coercive field, as well as the magnetization direction of the ferromagnetic overlayer, can be substantially tuned. These goals were achieved by the interfacial strain modification of the magnetic anisotropy energy profile. The observation and analysis of the electric field tunability of magnetization and the establishment of novel controlling schemes provide valuable directions for both theoretical development and future application endeavors.

Precisely measuring the flow rate in sprayers is a key technology to precision agriculture. With the development of advanced technologies, the demand for the ability to measure flow rate of individual nozzle has become more important and urgent. This paper investigates the possibility of developing a low-cost flow rate measurement technique. The technique is based on analyzing the acoustic signal from a microphone placed near the nozzle tip. A comparison between acoustic signal and vibration signal was made to study the relations between them. Then several possible locations of the microphone for measuring flow rate were tested and compared, and one has been chosen as the best location. After that, two methods of analyzing data were proposed, one that could better describe the original curve was chosen. With all of that work done, further experiments were conducted on a variety of nozzle tips. The results showed that an acoustic sensor could be used as an indicator of flow rate from a nozzle, but that unique calibrations for different nozzle tips would be necessary.

Distillers dried grains (DDGS), the coproducts of corn ethanol production, were used as a textile sizing agent on cotton, polyester and polyester/cotton blends in an effort to find inexpensive and biodegradable alternatives to sizing agents such as poly(vinyl alcohol) that are currently used. Although DDGS is an inexpensive, biodegradable and abundant co-product, it has limited industrial applications. DDGS is a mixture of carbohydrates, proteins and oil which are used as sizing agents or as size additives. The effects of DDGS extraction conditions on sizing evaluation parameters such as fiber adhesion, film properties, viscosity and fabric abrasion were studied in comparison with commercially available PVA or starch based size. Among the various conditions studied, DDGS that was extracted in 0.5% alkali concentration solution for 1 hour was most suitable for textiles sizing, based on the improvement in the strength and elongation of rovings and fabric abrasion resistance. The Chemical Oxygen Demand (COD) of DDGS size decreased 80% in 3 days compared to 25% for PVA demonstrating that DDGS was easily biodegradable in activated sludge. Although DDGS contains up to 30% protein, the total nitrogen and ammonia released from DDGS sizing agent had negligible effect on biodegradability. Good sizing performance and easy biodegradability demonstrate that DDGS based sizes have potential to serve as an environmentally friendly sizing agent.

Conventional enhanced biological phosphorus removal is based on anaerobic phosphorus release and aerobic phosphorus uptake, which requires more energy supply and sludge production. Thus, there are growing concerns about effective phosphorus removal from wastewater with lower energy requirement. The aim of this study was to achieve enriched sludge of denitrifying phosphorus accumulating organisms with anaerobic/anoxic conditions, especially with NO₂-N, and explore the possibility of anoxic EBPR process with NO₂-N as the sole electron acceptor in anaerobic/anoxic process. This study included an extensive literature review, laboratory works involving continuous operation of biological reactors and sample analysis for data collection, data analysis and process simulation for practical phosphorus removal, to address the research questions. The results suggested that long-period continuous NO₂-N dosing in anoxic phase could induce faster enrichment of denitrifying phosphorus accumulating organisms and more efficient phosphorus removal than NO₃-N at ambient temperature, without toxic inhibition of nitrite. The ratio of NO₃-N to NO₂-N was accorded to the amount of electron transfer of them, to remove the same amount of phosphorus. Dechloromonas is the functional microbial group in anoxic phosphorus uptake in both NO₂-N and NO₃-N based anaerobic/anoxic systems. It was concluded that the complete phosphorus removal could be achieved with NO₂-N as the sole electron acceptor in anaerobic/anoxic reactors, with enriched denitrifying phosphorus accumulating organisms.

Multidomain environments where multiple organizations interoperate with each other are becoming a reality as can be seen in emerging Internet-based enterprise applications. Access control to ensure secure interoperation in such an environment is a crucial challenge. A multidomain environment can be categorized as tightly-coupled and loosely-coupled. The access control challenges in the loosely-coupled environment have not been studied adequately in the literature. In a loosely-coupled environment, different domains do not know each other before they interoperate. Therefore, traditional approaches based on users' identities cannot be applied directly. Motivated by this, researchers have developed several attribute-based authorization approaches to dynamically build trust between previously unknown domains. However, these approaches all focus on building trust between individual requesting users and the resource providing domain. We demonstrate that such approaches are inefficient when the requests are issued by a set of users assigned to a functional role in the organization. Moreover, preserving principle of security has long been recognized as a challenging problem when facilitating interoperations. Existing research work has mainly focused on solving this problem only in a tightly-coupled environment where a global policy is used to preserve the principle of security. In this thesis, we propose a role-based access control and trust management framework for loosely-coupled environments. In particular, we allow the users to specify the interoperation requests in terms of requested permissions and propose several role mapping algorithms to map the requested permissions into roles in the resource providing domain. Then, we propose a Simplify algorithm to simplify the distributed proof procedures when a set of requests are issued according to the functions of some roles in the requesting domain. Our experiments show that our Simplify algorithm significantly simplifies such procedures when the total number of credentials in the environment is sufficiently large, which is quite common in practical applications. Finally, we propose a novel policy integration approach using the special semantics of hybrid role hierarchy to preserve the principle of security. At the end of this dissertation a brief discussion of implemented prototype of our framework is present.

Designing products for global markets with cultural backgrounds that differ from those of the company’s native culture (cross-cultural design) is an increasingly important challenge for many local and international companies. This thesis identifies different strategies and describes how companies can choose the best approach when working on cross-cultural design projects. Design is key to achieving business success in the international market place and maintaining harmonious cultural development. A person’s values are a combination of local/national cultural values and global cultural influences. The percentage varies by geography and product category. Cross-cultural design practitioners must better understand the balance when integrating local and global cultural factors. For this research three strategies incorporating culture research that contribute to the improvement of cross-cultural design projects are analyzed: 1) Global Brands, 2) Reconfiguring National Brands, and 3) Target Markets’ Culture Localizing International Brands. These strategies will enable cultural outsiders to design successfully for international markets.

Agents who perform interrelated tasks or work in similar local conditions often observe each other's actions and local signals. However, such information is often costly for the principal to obtain. Analytical models show that in such a situation, a peer reporting system with a verification mechanism (using one agent's information to verify the other's) and a reward for truthful whistle blowing can induce agents to report honestly and thereby help the principal achieve the first-best outcome. However, behavioral research suggests that the agents' perception regarding the fairness of the principal as well as cheap talk among agents may affect both how honestly agents report and how willing they are to blow the whistle on their peers. The results of the experiment show that under a peer reporting system, the agents' perception regarding the fairness of the principal positively affects the agents' reporting honesty and negatively affects the agents' rate of collusion. Communication between agents decreases their honesty and their whistle blowing when the principal is perceived as unfair, but not when the principal is perceived as fair.