137091 research outputs found
Sort by
The Influence of Action Observation on Sensorimotor Integration During Motor Performance
No full textSkill acquisition is a fundamental part of life for nearly every population, from babies learning to walk to persons in motor rehabilitation. It either entails the development of a new skill or the refinement of an existing one that was previously acquired. One way a skill can be learned is by watching or observing another individual perform the task. Action observation for motor learning involves viewing a motor task to replicate the action. There is evidence that action observation training, when combined with physical practice, is more effective than physical practice alone and presents unique opportunities to facilitate skill acquisition. Research on action observation has shown that action observation training benefits not only performance production variables, such as movement coordination patterns related to movement speed and interlimb relative coordination, but also performance outcome variables related to motor learning, including improved task performance. Action observation has been shown to improve the execution of movements due to the shared neural networks for observation and execution. While neural representations from action observation have been studied, the role of task difficulty on physiological outcomes, such as gaze behavior, kinematics, and perceived workload, remains unclear. The goal of the following studies is to understand the role of an action observation intervention on motor and visual systems under different levels of task difficulty.Ph.D.Applied Physiolog
A Custom Implementation of Clock Tree Synthesis for Monolithic 3D-IC
No full textThree-Dimensional Integrated Circuit (3D-IC) offer significant advantages over tradi-
tional planar Integrated Circuit (IC)s, such as higher performance, better power efficiency,
and higher device density, by stacking multiple layers of circuits vertically. A key tech-
nology enabling this is the Through-Silicon Via (TSV), which connects different layers but
also introduces design overheads and challenges like increased capacitance, thermal issues,
and fabrication complexity.
To attempt to mitigate these limitations, this project focuses on Monolithic Inter-tier Via
(MIV) a more advanced form of TSV, which allows for denser and higher-performance in-
tegration. Custom Electronic Design Automation (EDA) algorithms for 3D-IC, developed
alongside manufacturing advances, enable designers to maximize improvements in band-
width, design density, and power efficiency offered by new process technologies. As power
concerns continue to grow in industry, particularly for data centers and AI/ML accelerators,
optimizing hardware power in both performance-per-watt efficiency and overall power con-
sumption becomes increasingly critical.
Based on prior work, We have developed a functional algorithm that delivers zero-skew
clock trees for monolithic die-stacked Application Specific Integrated Circuit (ASIC)s, us-
ing the 3D Method-of-Means and Medians algorithm, with wire length minimized via the
Deferred-Merge Embedding algorithm. We hope to utilize this program as a framework,
to further optimize the clustering and MIV insertion functions of this algorithm through
machine learning and alternative buffering strategies in the future.M.S.Electrical and Computer Engineerin
Designing Organic Solvent Reverse Osmosis Membranes for The Separation of Liquid Hydrocarbons
No full textLiquid phase hydrocarbon separation accounts for 40-70% of both capital and operational costs in the chemical, pharmaceutical, food, and petroleum industries since most of these processes are heavily involved with organic solvents. As energy demand and associated carbon emission rises due to increasing global population, the development of large-scale sustainable separation methods with lower energy penalty and carbon footprint is essential to help mitigate the worse effects of energy and climate crisis. Membrane-based separations are considered a promising augment to conventional thermally driven based separations such as distillation due to inherently higher energy efficiency, lower carbon footprint and less space requirements. Polymeric membranes such as cellulose acetate have been extensively used in seawater reverse osmosis desalination process due to their excellent processability and high scalability, but would suffer from swelling, plasticization and dissolution when exposed to organic solvents. The low solvent stability of polymeric membranes serves as a major gap in applying highly scalable and processable polymer membranes to organic solvent environment separations.
This dissertation aims to fill this major gap and focuses on two main methods to engineer solution-processable materials with enhanced chemical stability, either through engineering existing materials or designing completely new materials. For engineering existing materials, vapor phase infiltration (VPI) is used as a post-treatment process for polymer materials after being cast into their membrane form. This one additional step transforms the pure polymer material into an organic-inorganic hybrid composite with enhanced properties, including solvent stability essential for organic solvent separations. For designing new materials, a new series of polymers is developed via the incorporation of fluorine-rich moieties in the aromatic polymer backbone to control the swelling of polymers, balancing practical separation productivity and separation efficiency of the membrane. Both strategies are discussed in detail, providing insights into developing next generation membrane fabrication methods for challenging organic solvent separation processes.Ph.D.Chemical and Biomolecular Engineerin
Computational Methods for Gene Regulatory Network Inference
No full textGene regulatory networks (GRNs) are commonly used to describe the complex regulatory relationships of transcription factors (TFs) and target genes. These networks are essential blueprints for a vast array of biological processes, including cellular development, response to stimuli, and disease progression. Despite their importance, accurate inference of GRNs from high-throughput data remains a fundamental challenge in bioinformatics due to dynamic and context-specific nature of cellular programs and the inherent noise in multi-modality, high-dimensional biological data.
This thesis presents a combination of computational approaches for GRN inference that address different aspects of these challenges. In the first project, we develop a novel simulator-supervised neural network framework, SPREd, for GRN reconstruction from transcriptomic data. This algorithm development project leverages synthetic data generation through a biophysics-based simulation model, allowing for training deep neural networks that directly predict the relationships between TFs and target genes. The approach we develop here offers an alternative to the established paradigm of training multi-variable models to predict a gene’s expression from the levels of TFs. Our simulator-supervised learning strategy addresses the common limitation of insufficient ground truth data in GRN inference by creating large-scale synthetic datasets that capture realistic regulatory dynamics. We test SPREd on diverse synthetic and real data sets, demonstrating its improved accuracy in identifying both direct and indirect regulatory relationships compared to other GRN inference models.
While the SPREd model enables us to learn the underlying regulatory logic from bulk gene expression profiles alone, the second project in this thesis focuses on the reconstruction of GRNs from single-cell multi-omics data in disease contexts. Specifically, we aim to identify key regulators driving atherogenesis through integrative analysis of scRNA-seq and scATAC-seq data from multiple time points and experimental conditions. We use a combination of cis-regulatory analysis and coexpression-based GRN inference to identify TFs that regulate atherogenesis-linked transcriptomic shifts in endothelial cells. We also develop a new method that incorporates cis-regulatory evidence as prior information in a neural network model of gene expression, which subsequently yields GRN edges through explainable machine learning techniques. We present a novel strategy for visualization of regulatory profiles of genes, by using supervised learning to map each gene to a low dimensional embedding, allowing a global view of all genes that captures their differential expression patterns and regulatory evidence.
Our systematic analysis provides insights into the regulatory programs controlling flow-induced reprogramming of endothelial cells during the development of atherosclerosis. We predict the TFs Creb3l2, Rela, and Mef2c to coordinate the transition from proatherogenic endothelial cells to pathological states, and test these predictions in vitro through collaboration.
Together, this thesis demonstrates that combining principled computational innovation with biological applications can overcome fundamental limitations in GRN inference, establishing a framework for more accurate regulatory network reconstruction and mechanistic understanding of disease processes.Ph.D.Bioinformatic
Hyper-gates Framework for Straight-line Programs and Their Applications in Nonlinear Algebra
No full textThe massive computations required by artificial intelligence and scientific computing have outpaced general-purpose CPUs, necessitating hardware acceleration via GPUs. Straight-Line Programs (SLPs) are an ideal solution as by definition they are branchless and do not consist of any conditionals, avoiding thread divergence, high memory consumption, and incorrect solutions due to guessing caused by branching on hardware. SLPs can enable an algorithm to be mapped as a predictable data-flow circuit, enabling massive parallelization. Nonlinear systems are foundational to science and engineering, with their solutions studied with algebraic geometry and approximated by numerical algebraic geometry algorithms like homotopy continuation. These solvers rely on automatic (auto) differentiation, and since auto differentiation can be represented as an SLP, it is of great interest to represent the entire solver as an SLP for hardware acceleration.
This thesis introduces HGates.m2, a framework that expands on SLPs by defining hyper-level gates to model nonlinear algorithms. We detail how autodifferentiation and an adaptation of the homotopy continuation predictor-corrector method are implemented within this framework. For several geometric problems, we demonstrate how to set up polynomial systems of equations using H-gates
Additive Manufacturing for RF Across Scales: From Inkjet Multiband RFID to Micro-Printed sub-THz Devices
No full textThis thesis explores how Additive Manufacturing (AM) can serve radio-frequency (RF) systems across two different applications. First, it presents well-known inkjet printing to fabricate a multi band Radio-Frequency Identification (RFID) tag with the goal of proving an innovative non-contiguous bandwidth integration algorithm for ranging. Second, it investigates novel micro-scale metal printing to enable sub-THz interconnects for compact multi-chip modules.
In the first part, a non-contiguous bandwidth integration approach for wireless sensing and ranging is introduced. Instead of relying on one wide, continuous band, the method uses information from several separated sub-bands and combines them in processing to emulate a wider effective bandwidth. To validate the idea, an inkjet-printed RFID tag has been designed, simulated, fabricated, and measured. Inkjet, although well-known, remains attractive here because it is accessible, repeatable, and fast. Alongside the hardware, it has been developed a signal-processing algorithm that fuses the non-contiguous sub-bands while minimizing the loss in ranging accuracy. The printed tag plus algorithm are evaluated through different ranging experiments.
In the second part, I shift to metal micro-printing, using Exaddon Ceres 3D printer, to address interconnects for future sub-THz systems. For instance, heterogeneous integration demands interconnects above 100 GHz. A custom efficient Voxelizer tool is introduced, enabling precise control of geometry at the micron scale. To validate its performance, some representative test structures are printed. Finally, fabrication of sub-THz interconnects with different materials is demonstrated
Vapor Phase Infiltration of Metal Halides into Conjugated Polymers for Doping and Photocatalytic Applications
No full textVapor phase infiltration (VPI) is a technique used to hybridize the bulk of a polymer by using a vapor phase precursor. The precursor adsorbs onto the surface, diffuses into the bulk and eventually reacts with the polymer. VPI has found various applications in membranes, energy storage, catalysis, etc. In this thesis I will show the application of VPI for the doping of conjugated polymers and creating hybrid conjugated polymer-metal oxide dye-sensitized photocatalysts. Conjugated polymers have a continuous chain of conjugation along the backbone giving them their sought after semiconducting properties. When these polymers are oxidized through a redox reaction, polaronic charge carriers are created increasing the overall conductivity.
Solution doping of conjugated polymers is well-documented and highly effective, but the organic solvents necessary to swell the polymer can also destroy any patterning or microstructure. The vapor phase precursors TiCl4 and VOCl3 are shown to effectively dope P3HT with an initially increasing conductivity with increasing exposure times that eventually begins to drop down. This same trend has been noted in literature. The initial rise in conductivity is confirmed to be an increase in number of charge carriers. However, the UV-Vis measurements both in situ and ex situ show that the decrease in conductivity is due to both scattering sites being generated and the polymer dedoping, a phenomenon that has not been reported previously. XPS analysis and depth profile data show how limiting the diffusion of the VOCl3 doping agent allows for the dedoping rate to exceed the doping rate beginning at the underside of the film.
Infiltrated MOx materials made by VPI have been previously seen as nothing but scattering agents decreasing the conductivity. Herein, I show that infiltrated TiOx clusters can be used as catalyst centers with the P3HT acting as a dye-sensitizing material. Photoluminescence spectra show that the visible light absorbed by the P3HT excites an electron which is transferred to the TiOx to perform photocatalysis. Furthermore, VPI is shown to make a superior architecture where the TiOx is concentrated towards the surface, as opposed to burying the MOx under the conjugated polymer as is common in literature
Flexible Space Junk Allocation & Waste Abatement (SPACE JAWA): Flexibility Framework to Screen Strategies & Options for Sustainable On-Orbit Servicing Infrastructures in LEO
No full textAs mega-constellations proliferate in Low Earth Orbit (LEO), the rate of satellite re-entry is accelerating dramatically. Between the start of this thesis and its defense, over 1,420 objects have re-entered Earth's atmosphere, with this rate expected to increase substantially over the next decade. While de-orbiting satellites mitigates orbital congestion, the environmental consequences of atmospheric re-entry remain uncertain, potentially impacting the ozone layer and climate. Transitioning from single-use satellites to a circular space economy through On-Orbit Servicing (OOS) could address these concerns by making space operations more sustainable.
Despite OOS success in Geosynchronous Orbit, LEO-based servicing remains economically challenging due to low launch costs and inefficient orbital maneuvering. Existing flexibility frameworks for OOS, while valuable for understanding customer demand, lack comprehensive treatment of multi-domain uncertainty, combinatorial flexible options, and novel concepts such as temporary orbital storage. This thesis makes three distinct contributions to address these gaps. First, it introduces Collection-as-a-Service (CAAS), a novel concept for circular space economies in LEO that enables satellite collection, refurbishment, and redeployment. Second, it develops a comprehensive flexibility framework that considers both satellite constellation and servicing infrastructure perspectives, modeling evolving on-orbit servicing space logistics through Monte Carlo scenarios with decision rules. Third, it integrates quantitative policy analysis into the flexibility framework, modeling various policy schemes within the discrete event simulation to assess their impact on total cost, NOx emission reduction, refurbishment rates, and OOS viability.
Through discrete event simulation and object-oriented programming, the framework models interactions between satellite constellations and servicing infrastructure across uncertain futures, incorporating multiple uncertain variables including launch costs, manufacturing costs, and technology obsolescence. Results demonstrate that specific architectural configurations achieve sustainability improvements while maintaining cost parity with traditional approaches. Configurations featuring one warehouse with rendezvous and proximity operations-capable satellites and flexible satellite serviceability upgrades provide the best balance of cost and emission reductions. Key enablers include dual-mission launch architectures for warehouse resupply and Earth-return for Earth-based refurbishment capabilities. The framework reveals that moderate policy interventions, such as a $50,000 Orbital Use Fee that forms a subsidy fund and enables rebates for satellite collections with a lump-sum subsidy of the remaining fund at the end of the 30-year simulation period, can achieve cost neutrality while improving refurbishment throughput.
Critically, no single element independently guarantees success across all metrics; rather, integrated consideration of technical configurations, flexible deployment strategies, and policy interventions is essential. These findings validate that the proposed flexibility framework is effective for identifying context-dependent strategies that balance economic viability, environmental sustainability, and operational resilience, providing policymakers and industry stakeholders with viable pathways toward circular space economies
Hexagonal boron nitride for deep UV applications
No full textCurrent UVC LEDs based on AlGaN technology face major challenges, primarily due to the drastic reduction in conductivity caused by the difficulty of achieving efficient p-type doping as the aluminum content increases. Additionally, light extraction efficiency decreases with higher aluminum content.
Hexagonal boron nitride (hBN) has emerged as a promising material to overcome these limitations. Owing to its unique properties, such as efficient p-type dopability, high UV transparency, strong luminescence, and excellent light extraction, hBN can potentially replace or complement conventional AlGaN layers. In particular, hBN can be integrated into AlGaN-based LEDs to form heterostructures with improved performance, simultaneously serving as both a hole injection layer and an electron blocking layer.
The objective of this thesis is to demonstrate the feasibility of such heterostructures by fabricating operational UV LEDs that incorporate hBN as a hole injection layer. Moreover, during this PhD, an extraordinary phenomenon, extremely long-lived persistent photoconductivity (PPC) in hBN at room temperature, has been discovered. This PPC effect, which can last for years, enables a novel photoinduced doping approach. Notably, the doping type can be tuned by varying the excitation wavelength. The understanding and utilization of this surprising effect constitute a central part of this work.
The results presented in this thesis represent a significant step toward efficient UVC emission using hBN/AlGaN heterostructures, providing a potential solution to the p-doping challenges at high aluminum content. Furthermore, the ability to optically induce both n-type and p-type doping in hBN opens the door to the realization of homojunction devices, paving the way toward entirely hBN-based LEDs with unprecedented efficiency, particularly since conventional doping techniques in hBN are still in their early stages