Now showing items 1-20 of 1250

    • Facade Segmentation in the Wild

      Para, Wamiq Reyaz (2019-08-19) [Thesis]
      Advisor: Wonka, Peter
      Committee members: Alouini, Mohammed-Slim; Thabet, Ali Kassem
      Facade parsing is a fundamental problem in urban modeling that forms the back- bone of a variety of tasks including procedural modeling, architectural analysis, urban reconstruction and quite often relies on semantic segmentation as the first step. With the shift to deep learning based approaches, existing small-scale datasets are the bot- tleneck for making further progress in fa ̧cade segmentation and consequently fa ̧cade parsing. In this thesis, we propose a new fa ̧cade image dataset for semantic segmenta- tion called PSV-22, which is the largest such dataset. We show that PSV-22 captures semantics of fa ̧cades better than existing datasets. Additionally, we propose three architectural modifications to current state of the art deep-learning based semantic segmentation architectures and show that these modifications improve performance on our dataset and already existing datasets. Our modifications are generalizable to a large variety of semantic segmentation nets, but are fa ̧cade-specific and employ heuris- tics which arise from the regular grid-like nature of fac ̧ades. Furthermore, results show that our proposed architecture modifications improve the performance compared to baseline models as well as specialized segmentation approaches on fa ̧cade datasets and are either close in, or improve performance on existing datasets. We show that deep models trained on existing data have a substantial performance reduction on our data, whereas models trained only on our data actually improve when evaluated on existing datasets. We intend to release the dataset publically in the future.
    • Development of Solution Processed Co-planar Nanogap Capacitors and Diodes for RF Applications Enabled Via Adhesion Lithography

      Felemban, Zainab (2019-08-18) [Thesis]
      Advisor: Anthopoulos, Thomas D.
      Committee members: Laquai, Frédéric; McCulloch, Iain
      Fabrication process of capacitors and Schottky diodes with nanogap electrodes is explained in this Thesis. The Schottky diode is made with IGZO in the nanogap, whereas the capacitor is made with ZrO2 in the nanogap which acts as the dielectric. Moreover, the electric characterization of both the diode and capacitor was obtained for different frequencies and different diameters. The end result showed that as the frequency increases the diode performance increases, but the capacitance of the capacitors decreases. Also, the barrier height and concentration were obtained using the Mott-Schottky plot for different frequencies. The 10MHz had the highest carrier concentration (5.9E+18cm-3) and barrier height (1V).
    • Expression of EZH1-Polycomb Repressive Complex 2 and MALAT1 lncRNA and their Combined Role in Epigenetic Adaptive Response

      Al Fuhaid, Lamya (2019-08-04) [Thesis]
      Advisor: Orlando, Valerio
      Committee members: Orlando, Valerio; Arold, Stefan T.; Al-Babili, Salim
      Living cells maintain stable transcriptional programs while exhibiting plasticity that allows them to respond to environmental stimuli. The Polycomb repressive complex 2 (PRC2) is a key regulator of chromatin structure that maintains gene silencing through the methylation of histone H3 on lysine 27 (H3K27me), establishing chromatin-based memory. Two variants of PRC2 are present in mammalian cells, PRC2-EZH2 which is predominantly present in differentiating cells, and PRC2-EZH1 that predominates in post-mitotic tissues. PRC2-EZH1α/β pathway is involved in the response of muscle cells to oxidative stress. Atrophied muscle cells respond to oxidative stress by enabling the nuclear translocation of EED and its assembly with EZH1α and SUZ12. Here we prove that the metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) long noncoding RNA (lncRNA) is required for the assembly of PRC2-EZH1 components. The absence of MALAT1 significantly decreased the association between EED and EZH1α proteins. Biochemical analysis shows that the presence of MALAT1 increases the enzymatic activity of PRC2-EZH1 in vitro. In addition, we show that the simultaneous expression of PRC2 core components is necessary for their solubility. The successful expression of PRC2 proteins enables the execution of several downstream experiments, which will further explain the nature of the interplay between MALAT1 and PRC2.
    • Variations in reef-associated fish communities in response to different benthic states in the east central Red Sea

      Short, George (2019-08) [Thesis]
      Advisor: Jones, Burton
      Committee members: Berumen, Michael; Carvalho, Susana
      Coral reefs are priority habitats which are vulnerable to natural and anthropogenic disturbances. These can cause phase shifts from coral habitat to degraded algal-dominated states – and consequent changes in the distribution, abundance and activity of associated fish species. In the eastern Red Sea, human-induced reef degradation is likely to increase with planned development of the Saudi Arabian coast and the changing climate. The present study therefore investigates the ecological effects of coral-algal phase shifts in reef-associated fish communities, using naturally occurring within-reef benthic zones as proxies for levels of habitat health - with a focus on how these responses differ temporally. These zones were dominated by: hard coral (coral zone), coral and turf algae (transition zone), and macroalgal canopies (algal zone). Six inshore reef areas, were studied in periods with low and high densities of Sargassum in the algal zones (May and November respectively). Community composition was assessed via visual census and predation activity predicted using two proxies: in situ experiments and biomass of carnivores. In both periods, we observed distinct fish communities in each zone - with reduced species richness, Shannon-Wiener diversity and predation intensity, from the coral to the algal zones. Decreases in the abundance and biomass of fish also occurred from the coral to algal zones in May but a spike, as well as a shift in community composition, occurred in the algal zone in November. This shift is attributed to the vast increases in grazer biomass, predominantly Siganus luridus, associated with the November bloom of Sargassum canopies. The present study established, the composition and functioning of Red Sea fish communities is spatially and temporally affected by increased macroalgal dominance. This finding supports the need for herbivorous fish to be made a conservation priority in the management and conservation of reef systems in order to prevent phase shifts to algal dominated states. We conclude that if Red Sea reefs are allowed to shift to alternate states, depending on the density of macroalgal canopies, reefs may support high biomass and abundance of fish but the functioning of the fish community will be altered and the diversity lost.
    • Analysis and Optimization of Massive MIMO Systems via Random Matrix Theory Approaches

      boukhedimi, ikram (2019-08-01) [Dissertation]
      Advisor: Alouini, Mohamed-Slim
      Committee members: Alouini, Mohamed-Slim; Laleg-Kirati, Taous-Meriem; Al-Naffouri, Tareq Y.; Kammoun,Alba; Li, Yonghui
      By endowing the base station with hundreds of antennas and relying on spatial multiplexing, massive multiple-input-multiple-output (MIMO) allows impressive advantages in many fronts. To reduce this promising technology to reality, thorough performance analysis has to be conducted. Along this line, this work is focused on the convenient high-dimensionality of massive MIMO's corresponding model. Indeed, the large number of antennas allows us to harness asymptotic results from Random Matrix Theory to provide accurate approximations of the main performance metrics. The derivations yield simple closed-form expressions that can be easily interpreted and manipulated in contrast to their alternative random equivalents. Accordingly, in this dissertation, we investigate and optimize the performance of massive MIMO in di erent contexts. First, we explore the spectral e ciency of massive MIMO in large-scale multi-tier heterogeneous networks that aim at network densi cation. This latter is epitomized by the joint implementation of massive MIMO and small cells to reap their bene ts. Our interest is on the design of coordinated beamforming that mitigates cross-tier interference. Thus, we propose a regularized SLNR-based precoding in which the regularization factor is used to allow better resilience to channel estimation errors. Second, we move to studying massive MIMO under Line-of-Sight (LoS) propagation conditions. To this end, we carry out an analysis of the uplink (UL) of a massive MIMO system with per-user channel correlation and Rician factor. We start by analyzing conventional processing schemes such as LMMSE and MRC under training-based imperfect-channel-estimates, and then, propose a statistical combining technique that is more suitable in LoS-prevailing environments. Finally, we look into the interplay between LoS and the fundamental limitation of massive MIMO systems, namely, pilot contamination. We propose to analyze and compare the performance using single-cell and multi-cell detection methods. In this regard, the single-cell schemes are shown to produce higher SEs as the LoS strengthens, yet remain hindered by LoS-induced interference and pilot contamination. In contrast, for multi-cell combining, we analytically demonstrate that M-MMSE outperforms both single-cell detectors by generating a capacity that scales linearly with the number of antennas, and is further enhanced with LoS.
    • Comparative metabolic modeling and analysis of human pathogens

      Abdel-Haleem, Alyaa M. (2019-08) [Thesis]
      Advisor: Gojobori, Takashi
      Committee members: Gao, Xin; Al-Babili, Salim; Bajic, Vladimir; Lewis, Nathan
      Infectious diseases continue to be major health concerns worldwide. Although major advances have led to accumulation of genomic data about human pathogens, there clearly exists a gap between genome information and studies aiming at identifying potential drug targets. Here, constraint-based modeling (CBM) was deployed to integrate disparate data types with genome-scale metabolic models (GEMs) to advance our understanding of the pathogenesis of infectious agents with respect to identifying and prioritizing drug targets. Specifically, genome-scale metabolic modeling of multiple stages and species of Plasmodium, the causative agent of malaria, was used to prioritize potential drug targets that could be used to simultaneously treat (anti-malarials) and block transmission of the parasite. In addition, species-specific metabolic models were used to guide translation of findings from non-human experimental disease models to human-infecting species. Further, comparative analysis of the essentiality of metabolic genes for V. cholerae, the causative agent of cholera, growth and survival in single and co-infections with other enteric pathogens led to prioritizing conditionally independent essential genes that would be potential drug targets in both single and co-infection scenarios. Taken together, our findings highlight the utility of using genome-scale metabolic models to prioritize druggable targets that would be of broader spectrum against human pathogens.
    • Evaluating the regulation of signaling pathways downstream of CD44 antibody treatment in AML

      Alghuneim, Arwa (2019-08) [Thesis]
      Advisor: Merzaban, Jasmeen
      Committee members: Liberale, Carlo; Blilou, Ikram
      Acute myeloid leukemia (AML) is a subset of leukemia that is characterized by the clonal expansion of cytogenetically and molecularly abnormal myeloid blasts. These blasts are highly proliferative accumulating in bone marrow and blood which leads to severe infections, anemia, and bone marrow failure. The poor prognosis of AML patients caused by the low tolerance to intensive chemotherapy has encouraged the pursuit of alternative therapeutic approaches. Differentiation therapy which involves the use of agents that can release the differentiation block in these leukemic blasts has emerged as a promising therapeutic approach. The use of All-trans retinoic acid (ATRA) represents a successful example of such an approach, nonetheless its efficacy is restricted to one subtype of AML. Efforts have been focused on finding differentiation agents which are effective for the other more common AML subtypes. Anti-CD44 targeted antibodies that activate the CD44 cell surface antigen are a promising candidate. Previous studies have shown that anti-CD44 treatment has been able to release the differentiation block in AML1 through AML5 subtypes. The exact mechanism by which anti-CD44 treatment is able to induce its effects has not been fully elucidated. Recent studies highlight the role that epigenetic mechanisms play during haematopoiesis and leukemogenesis and therefore, in this work we investigated the epigenetic mechanisms associated with anti-CD44 induced differentiation. Using AML cell lines from different subtypes, we demonstrated that anti-CD44-induced differentiation results in an extensive change of histone modification levels. We found that inhibiting enzymes responsible for the H3K9ac, H3K4me, H3K9me, and H3K27me modifications, attenuated the anti-proliferative and differentiation promoting effects of antic-CD44 treatment. Taken together, these data highlight the promising potential of using anti-CD44 as a therapeutic agent across multiple subtypes in AML
    • Study of ultraviolet AlGaN nanowires light-emitting diodes

      Priante, Davide (2019-08) [Dissertation]
      Advisor: Ooi, Boon S.
      Committee members: Ooi, Boon S.; Ohkawa, Kazuhiro; Schwingenschlögl, Udo; Mi,Zetian
      Ultraviolet (UV) group III-Nitride-based light emitters have been used in various applications such as water purification, medicine, lighting and chemical detection. Despite attractive properties such as bandgap tunability in the whole UV range (UV-C to UV-A), high chemical stability and relative low cost, the low quantum efficiency hamper the full utilization. In fact, external quantum efficiencies of UV devices are below 10 % for emission wavelength shorter than 350 nm. This thesis aims to show alternative solutions to such problems by employing nanowires (NWs) structures, and target the eventual application of reliable and high power NWs-based light-emitting devices, enabling large-scale production using the established silicon foundry processes. Here, we present the improvement of injection current and optical power of AlGaN NWs LEDs by involving a metal bilayer thin film with a dual purpose: eliminate the potential barrier for carrier transport, and inhibit the formation of silicide. We then study the AlGaN/GaN UV LED design to optimize the device structure and improve the LED performance. We compared multiple devices having different active region and graded layers’ thicknesses. Improvement on the output power was achieved for larger p-AlGaN graded layer and thinner p-GaN contact layer structure due to the better hole injection and lower p-GaN absorption. The junction temperature of AlGaN-based NWs LEDs on metal bi-layer and silicon is also presented as a crucial parameter affecting the device efficiency, chromaticity and reliability. In this regard, by using the forward-voltage and peak-shift method we extracted the junction temperature values and confirmed the better heat dissipation in NWs grown on metal substrate. Finally, the origin of single and ensemble NWs current injection and injection efficiency are studied by treating the AlGaN NWs with KOH solution. Measurements based on conductive atomic force microscopy enabled a fast feedback cycle without fabricating the device. Despite the NWs technology is still at its infancy compared to the matured planar, we believe that this research effort will give important insight in advancing the AlGaN NWs devices for future industrial employment.
    • Design and Real-time Implementation of Model-free Control for Solar Collector

      Alharbi, Mohammad (2019-08) [Thesis]
      Advisor: Laleg-Kirati, Taous-Meriem
      Committee members: Ahmed, Shehab; Kammoun,Alba; Diagne, Mamadou
      This work addresses the design and real-time implementation of adaptive control strategies on the parabolic solar collector to enhance the production efficiency under varying working conditions. For example, the unpredictable variations of the solar irradiance and thermal losses, these factors can be a major problem in the control design. The control objective is to force the outlet temperature of the collector fluid, to track a predefined reference temperature regardless of the environmental changes. In this work, two control strategies have been designed and analyzed. First, an intelligent proportional-integral feedback control, which combines the proportionalintegral feedback control with an ultra-local model is proposed. This strategy uses a transfer function model that has been derived and identified from real-time data and used to test the controller performance. Second, an adaptive nonlinear control using Lyapunov stability theory combined with the phenomenological representation of the system is introduced. This strategy uses a bilinear model derived from the heat transfer equation. Both control strategies showed good performance in the simulations with respect to the convergence time and tracking accuracy. Besides, the conventional proportional-integral controller has been successfully implemented in the real system.
    • SPECTRUM MANAGEMENT FOR FUTURE GENERATIONS OF CELLULAR NETWORKS

      Randrianantenaina, Itsikiantsoa (2019-08) [Dissertation]
      Advisor: Alouini, Mohamed-Slim
      Committee members: Alouini, Mohamed-Slim; Ghanem, Bernard; Shihada, Basem; Dahrouj, Hayssam; Bennis, Mehdi
      The demand for wireless communication is ceaselessly increasing in terms of the number of subscribers and services. Future generations of cellular networks are expected to allow not only humans but also machines to be immersively connected. However, the radio frequency spectrum is already fully allocated. Therefore, developing techniques to increase spectrum efficiency has become necessary. This dissertation analyzes two spectrum sharing techniques that enable efficient utilization of the available radio resources in cellular networks. The first technique, called full-duplex (FD) communication, uses the same spectrum to transmit and receive simultaneously. Using stochastic geometry tools, we derive a closed-form expression of an upper-bound for the maximum achievable uplink ergodic rate in FD cellular networks. We show that the uplink transmission is vulnerable to the new interference introduced by FD communications (interference from the downlink transmission in other cells), especially when the disparity in transmission power between the uplink and downlink is considerable. We further show that adjusting the uplink transmission power according to the interference power level and the channel gain can improve the uplink performance in full-duplex cellular networks. Moreover, we propose an interference management technique that allows a flexible overlap between the spectra occupied by the downlink and uplink transmissions. The flexible overlap is optimized along with the user-to-base station association, the power allocation and the channel allocation in order to maximize a network-wide utility function. The second spectrum sharing technique, called non-orthogonal multiple access (NOMA), allows a transmitter to communicate with multiple receivers through the same frequency-time resource unit. We analyze the implementation of such a scheme in the downlink of cellular networks, more precisely, in the downlink of fog radio access networks (FogRANs). FogRAN is a network architecture that takes full advantage of the edge devices capability to process and store data. We propose managing the interference for NOMA-based FogRAN to improve the network performance by jointly optimizing user scheduling, the power allocated to each resource block and the division of power between the multiplexed users. The simulation results show that significant performance gains can be achieved through proper resource allocation with both studied spectrum sharing techniques.
    • Tra c Monitoring and MAC-Layer Design for Future IoT Systems

      Odat, Enas M. (2019-08) [Dissertation]
      Advisor: Shamma, Jeff S.
      Committee members: Shihada, Basem; Al-Naffouri, Tareq Y.; Pavel, Lacra
      The advances in the technology and the emergence of low complexity intelligent devices result in the evolution of the Internet-of-Things (IoT). In most IoT application scenarios, billions of things are interconnected together using standard communication protocols to provide services for di erent applications in the healthcare industry, smart cities, transportation, and food supply chain. Despite their advantage of connecting things anywhere, anytime, and anyplace, IoT presents many challenges due to the heterogeneity, density, the power constraints of things, and the dynamic nature of the network that things might connect and disconnect at any time. All of these increase the communication delay and the generated data, and it is thereby necessary to develop resource management solutions for the applications in IoT. One of the most important resources is the wireless channel, which is a shared resource; thus, it is necessary for the nodes to have methods that schedule channel access. This thesis considers the problem of distributed sensing and channel access in the context of IoT systems, where a set of sel sh nodes competes for transmission opportunities. In the channel access part, a memory-one channel access game is proposed to reduce the collision rate, to enhance the cooperation among the nodes, and to maximize their payo s by optimizing their channel access probabilities, based on the channel state in the previous time step. To overcome the communication cost overhead in the network and to solve the problem e ciently, the nodes use distributed learning algorithms. Next, the problem is extended to include energy constraints on the transmission decisions of the nodes, where each one of them has a battery of nite capacity, which is replenished by an energy-harvesting process. This constrained problem is solved using energy-aware channel access games under di erent scenarios of perfect and imperfect information. In the distributed sensing part, a tra c-monitoring system, integrated into a WSN, is proposed as a potential application to implement the channel access solution. This system maximizes the privacy of the sensed tra c by using low-cost and low-power sensor devices that integrate passive infrared sensors (PIR) and ultrasonic range nders. To estimate the parameters required to solve the real-time monitoring problem (vehicle detection, classi cation, and speed estimation), the measurements of these sensors are analyzed using a set of optimized machine-learning algorithms. The selection of these algorithms is due to the continuous variation of the sensed environment over time, the lack of the system state dynamic models, and the limitation in the resources.
    • Optimizing a Selective Whole Genome Amplification (SWGA) Strategy for Clinical Malaria Infections

      Alawi, Mariah (2019-08) [Thesis]
      Advisor: Pain, Arnab
      Committee members: Habuchi, Satoshi; Blilou, Ikram
      Plasmodium is a genus well known for causing malaria, a life-threatening infection for many people where malaria is endemic. The blood-borne disease is transmitted by the female Anopheles mosquito. Till date, eight parasite species have been reported to cause malaria in humans that include P. falciparum, P. vivax, P. malariae, P. ovale curtisi, P. ovale wallikeri, P. cynomolgi, P. knowlesi and more recently P. simium. Amongst them, the most genetically understood species is P. falciparum, causing most of the deaths in children from malaria. Understanding genome variation at the population level of all malaria species is of utmost importance, including clinical cases with very low parasitemia. To achieve this purpose, we need sufficient amounts of parasite DNA material from the pool of host DNA, which always is overrepresented in clinical infections. We utilized a strategy of selective whole genome amplification (SWGA) technology on P. malariae and P. ovale curtisi (two neglected human infecting malaria parasites that often cause mild yet clinically relevant infections with low parasitemia) to efficiently enrich their genomic DNA for high-quality whole genome sequencing. Previous studies on SWGA applied on P. falciparum and P. vivax showed that SWGA could efficiently enrich the amount of starting DNA material from inadequate amounts of parasites directly from clinical samples without separating the host DNA using specifically designed primer sets. We have successfully designed multiple sets of primers and tested the efficiency of five best primer sets using polymerase chain reaction to enrich the genomes of P. malariae and P. ovale curtisi. The efficiency of primers in enriching the genome was tested on two clinical samples for each of P. malariae and P. ovale curtisi. We were able to enrich the genome of P. malariae with an average of 19-fold (19X) enrichment across both samples. For P. ovale curtisi, we could achieve an enrichment of 3 folds only. Nevertheless, we still obtained a sufficient amount of gDNA to prepare Illumina sequencing libraries and call for SNPs and Indels in a biologically reproducible manner at genome-scale.
    • Identification of MALAT1 as a PRC2-Ezh1 Associated lncRNA Essential for Epigenetic Control of Skeletal Muscle Adaptation and Plasticity

      El Said, Nadine H. (2019-08) [Dissertation]
      Advisor: Orlando, Valerio
      Committee members: Froekjaer Jensen, Christian; Tester, Mark; Rinn, John
      Polycomb Proteins (PcG) are chromatin proteins that control the maintenance of “transcriptional memory” and cell identity by fixing the repressed state of developmentally regulated genes. This function has been linked to interaction with RNA moieties, in particular long non-coding RNAs (lncRNAs). However, specificity of PcG-RNA interactions has been controversial (Beltran et al., 2016; Chen Davidovich, Leon Zheng, Karen J. Goodrich, & Thomas R. Cech, 2013). In this study we took advantage of recent work published from our lab reporting about a novel and reversible mechanism regulating genome wide Ezh1-PRC2 activation in mouse skeletal muscle cells in response to atrophic stress (Bodega et al., 2017). Using this physiological, in vivo tool we could identify a functional dynamic crosstalk between Malat1 (Metastasis Associated Lung Adenocarcinoma Transcript 1) and PRC2-Ezh1 complex. By combining immuno-fluorescence, biochemistry, epigenomics, ChIRP, DNA and RNA immunoprecipitation we identified a novel pathway in which Malat1 plays a role in compartmentalization, assembly and activity of PRC2 in chromatin, allowing epigenetic plastic response to atrophic stress and recovery. We conclude that Malat1 is an essential partner for PRC2-Ezh1 adaptive function in skeletal muscle cells.
    • Designing Surfaces for Enhanced Water Condensation and Evaporation

      Jin, Yong (2019-08) [Dissertation]
      Advisor: Wang, Peng
      Committee members: Nunes, Suzana; Lai, Zhiping; Wang, Zuankai
      With the increasing pressure of providing reliable potable water in a sustainable way, it is important to understand water phase change phenomena (condensation and evaporation) as the water phase change is involved in many processes such as membrane distillation and solar still which can be a feasible choice of supplementing the present potable water access. In the present thesis, we first elucidate the role of wettability of water condensation substrate by combining the droplet growth dynamics and droplet population evolution. The results show that wettability has a negligible effect on water condensation rate in an atmospheric environment. After confirming the role of substrate wettability, we provide a quantitative analysis of the effect of substrate geometry on water condensation in the atmospheric environment. The analysis can help to predict the efficiency of water condensation rate with a given substrate of a certain geometry with the aid of computational simulation tools. The results show that water condensation can be increased by 40% by rationally designing the geometry of the condensation surface. However, the condensation rate in the atmospheric environment is relatively slow due to the presence of non-condensable gas. In order to increase the condensation rate, a relatively pure vapor environment is desired, in which condensed water will be the major heat transfer barrier. Coalescence induced jumping of condensed droplets on superhydrophobic surfaces is an interesting phenomenon to help faster removal of condensed droplets. However, it is still not clear how to optimize the overall heat transfer efficiency by condensation on such surfaces. We observed an interesting phenomenon on a superhydrophobic nano-cones array, on which water preferentially condenses within larger cavities among the nanocones. Droplets growing form larger cavities have larger growth rate. 6 This finding can possibly provide a solution to optimizing heat transfer efficiency. Finally, a nylon-carbon black composite is prepared by electrospinning to enhance water evaporation under solar radiation. The composite shows an interesting light absorption property. In a wet state, the composite can absorb around 94% of the incident sunlight. The composite also shows strong mechanical and chemical stability. Thus, the composite is considered to be a practical candidate to be applied in the solar distillation process.
    • Single-Crystal Halide Perovskites for High Efficiency Photovoltaics

      Alsalloum, Abdullah Yousef (2019-07-27) [Thesis]
      Advisor: Bakr, Osman
      Committee members: Al-Shareef, Reem A.; Mohammed, Omar F.
      Lead halide perovskite solar cells (PSCs) are considered the fastest growing photovoltaic technology, reaching an outstanding certified power conversion efficiency of 24.2% in just 10 years. The best performing PSCs are based on polycrystalline films, where the presence of grain boundaries and ultra-fast crystallization limit the further development of their performance by increasing the bulk and surface defects. Compared to their polycrystalline counterparts, single crystals of lead halide perovskites have been shown to possess much lower trap-state densities and diffusion lengths exceeding 100𝜇�������m. In this thesis, using a solution space-limited inverse temperature crystallization method, twenty-microns thick single crystals of MAPbI3 are grown directly on the charge selective contact to construct highly reproducible p-i-n inverted type solar cells with fill factors(FF) as high as 84.3% and power conversion efficiencies (PCEs) exceeding 21% under 1 sun illumination (AM 1.5G). A key requisite for high PCEs is avoiding surface hydration, in which moisture attacks the perovskite/transporting layer interface and causes a significant decrease in short-circuit current. These solar cells set a record for single crystal PSCs, and highlight the potential of single crystal PSCs in furthering perovskite photovoltaic technology.
    • Studying Perturbations on the Input of Two-Layer Neural Networks with ReLU Activation

      Alsubaihi, Salman (2019-07) [Thesis]
      Advisor: Ghanem, Bernard
      Committee members: Al-Naffouri, Tareq Y.; Thabet, Ali Kassem
      Neural networks was shown to be very susceptible to small and imperceptible perturbations on its input. In this thesis, we study perturbations on two-layer piecewise linear networks. Such studies are essential in training neural networks that are robust to noisy input. One type of perturbations we consider is `1 norm bounded perturbations. Training Deep Neural Networks (DNNs) that are robust to norm bounded perturbations, or adversarial attacks, remains an elusive problem. While verification based methods are generally too expensive to robustly train large networks, it was demonstrated in [1] that bounded input intervals can be inexpensively propagated per layer through large networks. This interval bound propagation (IBP) approach lead to high robustness and was the first to be employed on large networks. However, due to the very loose nature of the IBP bounds, particularly for large networks, the required training procedure is complex and involved. In this work, we closely examine the bounds of a block of layers composed of an affine layer followed by a ReLU nonlinearity followed by another affine layer. In doing so, we propose probabilistic bounds, true bounds with overwhelming probability, that are provably tighter than IBP bounds in expectation. We then extend this result to deeper networks through blockwise propagation and show that we can achieve orders of magnitudes tighter bounds compared to IBP. With such tight bounds, we demonstrate that a simple standard training procedure can achieve the best robustness-accuracy tradeoff across several architectures on both MNIST and CIFAR10. We, also, consider Gaussian perturbations, where we build on a previous work that derives the first and second output moments of a two-layer piecewise linear network [2]. In this work, we derive an exact expression for the second moment, by dropping the zero mean assumption in [2].
    • Asymptotic Performance Analysis of the Randomly-Projected RLDA Ensemble Classi er

      Niyazi, Lama (2019-07) [Thesis]
      Advisor: Alouini, Mohamed-Slim
      Committee members: Al-Naffouri, Tareq Y.; Kammoun,Alba; Dahrouj, Hayssam
      Reliability and computational e ciency of classi cation error estimators are critical factors in classi er design. In a high-dimensional data setting where data is scarce, the conventional method of error estimation, cross-validation, can be very computationally expensive. In this thesis, we consider a particular discriminant analysis type classi er, the Randomly-Projected RLDA ensemble classi er, which operates under the assumption of such a `small sample' regime. We conduct an asymptotic study of the generalization error of this classi er under this regime, which necessitates the use of tools from the eld of random matrix theory. The main outcome of this study is a deterministic function of the true statistics of the data and the problem dimension that approximates the generalization error well for large enough dimensions. This is demonstrated by simulation on synthetic data. The main advantage of this approach is that it is computationally e cient. It also constitutes a major step towards the construction of a consistent estimator of the error that depends on the training data and not the true statistics, and so can be applied to real data. An analogous quantity for the Randomly-Projected LDA ensemble classi er, which appears in the literature and is a special case of the former, is also derived. We motivate its use for tuning the parameter of this classi er by simulation on synthetic data.
    • Consequences of Coral-Algal Phase Shifts for Tropical Reef Ecosystem Functioning

      Roth, Florian (2019-07) [Dissertation]
      Advisor: Jones, Burton
      Committee members: Morán, Xosé Anxelu G.; Haas, Andreas; Wild, Christian; Daffonchio, Daniele; Carvalho, Susana
      Tropical coral reefs provide important ecosystem goods and services that are supported by one or more ecosystem functions (e.g., recruitment, primary production, calcification, and nutrient recycling). Scleractinian corals drive most of these functions, but a combination of global and local anthropogenic stressors has caused persistent shifts from coral- to algae-dominated benthic reef communities globally. Such phase shifts likely have major consequences for ecosystem functions; yet, related knowledge is scarce in general, but particularly at the community level, under ‘in situ’ conditions, and under the influence of changing environmental variables. Thus, we conducted a series of interconnected in situ experiments in coral- and algae-dominated reef communities in the central Red Sea, combining traditional community ecology approaches with novel metabolic and biogeochemical assessments from December 2016 to January 2018. Specifically, we (i) examined the influence of coral-algal phase shifts on recruitment and succession patterns, (ii) assessed the role of benthic pioneer communities in reef carbon and nitrogen dynamics, (iii) developed a novel approach to measure functions of structurally complex reef communities in situ, and (iv) quantified biogeochemical functions of mature coral- and algae-dominated reef communities. The findings suggest that coral-algal phase shifts fundamentally modify critical reef functions at different levels of biological organization, namely from pioneer to mature reef communities. For example, community shifts, through a lower habitat complexity and grazing pressure, decreased the number of coral recruits by >50 %, thereby inhibiting the replenishment of adult coral populations. At the same time, a 30 % higher productivity (annual mean) and increased organic carbon retention in algae-dominated communities supported a fast biomass accumulation and community growth, altering the habitat-specific community metabolism and reef biogeochemistry. Seasonal warming amplified these functional differences between coral- and algae-dominated communities, likely promoting a positive feedback loop of reef degradation under predicted ocean warming. Overall, this dissertation provides quantitative data on critical functions of classical and phase shifted novel reef communities, on tipping points for the collapse of community functions, and potential future winners and losers. The knowledge gained with this thesis helps, thereby, to understand how phase-shifted reef ecosystems function and which services will be generated in comparison to coral-dominated reefs under near-future stress scenarios.
    • Coral Bleaching – Breakdown of a Nutrient Exchange Symbiosis

      Rädecker, Nils (2019-07) [Dissertation]
      Advisor: Voolstra, Christian R.
      Committee members: Wild, Christian; Bosch, Thomas C.G.; Duarte, Carlos M.; Hirt, Heribert; Bosch, Thomas C.G.
      For millions of years, the nutrient exchange symbiosis between corals and their endosymbiotic algae has formed the foundation of the ecological success of coral reefs. Yet, in recent decades anthropogenic climate change is increasingly destabilizing this symbiosis, and thus the reefs that rely on it. High-temperature anomalies have caused mass mortality of corals due to repeated coral bleaching, the expulsion or digestion of symbionts by the host during stress. Hence, in-depth knowledge of the cellular processes of bleaching is required to conceive strategies to maintain the ecological functioning of coral reefs. In this thesis, we investigated the role of symbiotic nutrient cycling in the bleaching response of corals. For this, we examined the mechanisms that underlie the functioning of the symbiosis in a stable state and how heat stress affects these metabolic interactions during coral bleaching. Our findings reveal that the functioning of the coral – algae symbiosis depends on the resource competition between host and symbionts. In a stable state, symbiotic competition for ammonium limits nitrogen availability for the algal symbiont, thereby ensuring symbiotic carbon translocation and recycling. During heat stress, however, increased metabolic energy demand shifts host metabolism from amino acid synthesis to degradation. The resulting net release of ammonium by the host, coupled with the stimulated activity of associated nitrogen-fixing microbes, substantially increases nitrogen availability for algal symbionts. Subsequently, stimulated algal growth causes selfish retention of carbon, thereby further reducing energy availability for the host. This positive feedback loop disturbs symbiotic nutrient recycling, eventually causing the collapse of carbon translocation by the symbiont. Hence, heat stress causes shifts in metabolic interactions, which directly and indirectly destabilizes the symbiosis, and ultimately undermines the ecological benefits of hosting algal symbionts for corals. In summary, this thesis shows that integrating symbiotic nutrient cycling into our conceptual understanding of coral bleaching is likely to improve our ability to predict coral bleaching in light of environmental conditions and may ultimately help to conceive new strategies to preserve coral reef functioning.
    • Interactive Exploration of Objective Vortex Structures in Unsteady Flow using Observer Fields

      Shaker, Ghofran H. (2019-07) [Thesis]
      Advisor: Hadwiger, Markus
      Committee members: Hadwiger, Markus; Wonka, Peter; Moshkov, Mikhail
      Successful characterization of vortex structures in unsteady flow fields depend crucially upon an adequate choice of a reference frame. Vortex detection approaches in flow visualization aspire to be objective, i.e., invariant under time-dependent rotations and translations of the input reference frame. However, objectivity by itself does not guarantee good results as different specific approaches lead to different results. Moreover, recent more generic approaches to objectivity still require parameters to be specified beforehand which can significantly influence the resulting vortex detection, depending on the complexity and characteristics of the input flow field. With the assumption that human intervention is unavoidable to some extent, we tackle the problem of specifying parameters for vortex detection from a human-centered perspective. In this work, we present a novel system that enables users to interactively explore the parameter space of a flexible objective method, while jointly computing and visualizing the resulting vortex structures. We build on the computation of an objective field of reference frames and enable users to interactively change computation parameters as well as choose different observers, compute vortex structures on-the-fly during exploration, and visualize the flow field from the viewpoint of the chosen observers. Overall, we illustrate that such an interactive approach can be of significant value to the user for analyzing vortex structures visually and understanding why a computational method has detected a specific structure as a vortex.