12023 research outputs found
Sort by
Methods for Robust Learning-Based Control
Get PDFThis thesis addresses the general problem of improving control, safety, and reliability of multi-rotor drones in various challenging conditions by introducing novel deep-learning-based approaches. These approaches are designed to tackle specific issues that multi-rotor drones face during operation, such as near-ground trajectory control, high-speed wind disturbances, actuation delays, and motor failures. The thesis is organized into four main chapters, plus an introduction and conclusion. Each of the main chapters focuses on a unique approach to address a particular challenge of deep-learning-based control methods. Chapter 2 presents Neural-Lander, a deep-learning-based robust nonlinear controller that significantly improves quadrotor control performance during landing by accounting for complex aerodynamic effects. This chapter addresses key challenges to incorporating learned residual dynamics into a control architecture, laying the groundwork for the subsequent chapters. Chapters 3 and 4 introduce Neural-Fly, a learning-based approach that uses Domain Adversarially Invariant Meta-Learning (DAIML) and adaptive control to enable rapid online learning and precise flight control under a wide range of wind conditions. Chapter 5 proposes a lightweight augmentation method that enhances trajectory tracking performance for UAVs by effectively compensating for motor dynamics and digital transport delays. This method is extensible to a range of control methods, including learning-based approaches. Chapter 6 explores a novel sparse failure identification method for detecting and compensating for motor failures in over-actuated UAVs, contributing to the development of robust fault detection and compensation strategies for a safer and more reliable operation. This method builds on the Neural-Fly online learning framework and extends it to handle a wider range of conditions, including complete actuator failures. Together, these chapters address key challenges in safe and reliable learning-based control and demonstrate the potential of deep-learning-based control methods.</p
Application of Dewar Heterocycles and Vinyl Carbocations in Organic Synthesis
Get PDFHigh-energy molecules are frequently employed in the construction of organic molecules and materials in both academic and industrial settings. This thesis describes the application of two distinct classes of reactive molecules in organic synthesis: (1) Dewar heterocycles, which contain a highly strained bicyclic structure; (2) vinyl cations, a class of dicoordinated carbocations containing an electron-deficient carbon center bound to only two atoms. A description of our experimental work relevant to this thesis commences with the exploration of Dewar pyrone in the total synthesis of (±)-vibralactone. Next, the application of Dewar heterocycles to the synthesis of new strained-ring polymers will be discussed, including examples of post-polymerization strategies to access soluble poly(acetylene) derivatives and β-amino acid type polymers. Finally, the development of a catalytic asymmetric C–H insertion reaction of vinyl carbocations will be described, with an emphasis on reaction development, scope, and mechanism.</p
New HCR Technologies: 10-Plex Quantitative Spectral Imaging of RNAs and Proteins; Multiplexed Quantitative Imaging of Protein:Protein Complexes; and Sensitive, Instrument-Free, At-Home Pathogen Detection
Get PDFSignal amplification based on the mechanism of hybridization chain reaction (HCR) enables researchers to quantitatively image RNA and protein expression in highly autofluorescent biological samples. This thesis extends the capabilities of HCR to three new domains: spectral HCR imaging for quantitative 10-plex immunofluorescence and in situ hybridization in highly autofluorescent samples; imaging of protein:protein complexes using cooperative probes for logical control over HCR signal amplification; and HCR lateral flow tests for sensitive, instrument-free, at-home testing for infectious diseases.
While 4- or 5-plex imaging is readily achieved using orthogonal HCR systems labeled with spectrally distinct fluorophores, higher levels of multiplexing are challenging due to overlap in the broad excitation and emission spectra of commonly used fluorophores. In Chapter 2, we simultaneously image a combination of 10 protein and RNA targets via spectral imaging with linear unmixing. A combination of 10 reference spectra for 10 fluorophores chosen for optimal unmixing, 10 orthogonal HCR systems, and 11 optimized excitation and emission settings enable robust, user-friendly performance, which is demonstrated in whole-mount zebrafish embryos and mouse brain sections. We validate that unmixed subcellular voxel intensities enable accurate and precise relative target quantitation with subcellular resolution across all 10 channels and demonstrate single-molecule sensitivity and resolution for absolute RNA quantitation.
In Chapter 3, we introduce an enzyme-free method for multiplexed imaging of protein:protein complexes using split-initiator HCR signal amplification. Antibodies specific to each protein of the complex carry fractional initiators that become colocalized upon introduction of a DNA ruler strand to form a full HCR initiator and trigger growth of a tethered amplification polymer. Automatic background suppression is present throughout the protocol, as split-initiator antibody probes that bind to the sample nonspecifically or to isolated protein targets are too far apart to become colocalized by the ruler strand, precluding colocalization of a full initiator and preventing HCR signal amplification. We demonstrate the technique with high signal-to-background in adherent mammalian cells, pro-T cells, and highly autofluorescent formalin-fixed paraffin-embedded human breast tissue sections. Leveraging existing orthogonal HCR amplifiers, we design three orthogonal cooperative junctions for simultaneous 3-plex detection of protein:protein complexes. We validate that quantitative subcellular voxel intensities are generated, allowing for built-in relative quantitation of protein:protein complexes within the spatial context of the sample. Lastly, we demonstrate simultaneous detection of protein targets, RNA targets, and protein:protein complexes via a unified protocol for HCR immunofluorescence, in situ hybridization, and protein:protein complex imaging.
In Chapter 4, we enhance the sensitivity of conventional unamplified lateral flow tests for at-home infectious disease testing by developing an amplified assay with isothermal, enzyme-free signal amplification based on the mechanism of HCR. Traditional lateral flow tests are amenable to at-home testing and return a result within 10–15 minutes but demonstrate a high false-negative rate (e.g., 25-50% for SARS-CoV-2) due to the absence of signal amplification. The HCR lateral flow assay we develop maintains the simplicity of the conventional lateral flow assay user experience via a disposable 3-channel lateral flow device to automatically deliver reagents to the test region in three successive stages without user interaction. To perform a test, the user loads the sample, closes the device, and reads the result by eye after 60 minutes. Detecting gamma-irradiated SARS-CoV-2 virions in a mixture of saliva and extraction buffer, the current amplified HCR lateral flow assay achieves a limit of detection of 200 copies/μL using available antibodies to target the SARS-CoV-2 nucleocapsid protein. By comparison, five commercial unamplified lateral flow assays that use proprietary antibodies exhibit limits of detection of 500 copies/μL, 1000 copies/μL, 2000 copies/μL, 2000 copies/μL, and 20,000 copies/μL. By swapping out antibody probes to target different pathogens, amplified HCR lateral flow assays offer a platform for simple, rapid, and sensitive at-home testing for infectious diseases.</p
Bioorthogonal Noncanonical Amino Acid Tagging for Understanding Bacterial Persistence
Get PDFPhenotypic heterogeneity in populations of isogenic bacterial cells includes variations in metabolic rates and responses to antibiotic treatment. In particular, sub-populations of “persister” cells exhibit increased antibiotic tolerance. Understanding the mechanisms that underlie bacterial persistence would constitute an important step toward preventing and treating chronic infections. On the other hand, bacteria often have multiple molecular mechanisms to adapt to fluctuating environments. Understanding these mechanisms, and their redundancy, requires examinations in depth at the molecular level. This thesis describes a time- and cell state-selective proteome-labeling approach that enables researchers to investigate heterogeneous systems and molecular redundancy.
In Chapter 1, we review the concept of bacterial persistence. The definition of bacterial persistence is introduced. Both the differences and connections between bacterial persistence and resistance are covered. In particular, we discuss research related to Pseudomonas aeruginosa (P. aeruginosa), an important opportunistic pathogen found in many cystic fibrosis patients. State-of-the-art technologies to investigate bacterial persistence are discussed, and we conclude that advanced tools are needed to advance research on bacterial persistence further.
In Chapter 2, we highlight the concept of bioorthogonal noncanonical amino acid tagging (BONCAT). BONCAT is a powerful tool developed in the Tirrell and Schuman laboratories allowing the incorporation of noncanonical amino acids (ncAA) into newly-synthesized proteins. We review established strategies for proteomics, especially cell-selective proteomics. We introduce the concept and mechanism of BONCAT and address the advantages of BONCAT in the investigation of phenotypic heterogeneity and bacterial persistence.
In Chapter 3, we describe our work using BONCAT for understanding bacterial persistence. In particular, we investigated the process of persister resuscitation, as it is closely related to the reoccurrence of P. aeruginosa infections. The characteristics of the heterogeneity of persister cells during persister awakening were examined by survival assays and by ScanLag, an automated colony-based system allowing high-throughput acquisition of time-lapse images, quantification, and analysis of growth of bacterial colonies. Two BONCAT methods were developed in the P. aeruginosa strain PA14 by treating cells either with L-azidohomoalanine (Aha), which avoids extensive usage of antibiotic markers and allows direct integration with PA14 transposon insertion library, or with L-azidonorleucine (Anl), which has the advantage of specificity, as well as direct application in nutrition-rich medium. Through BONCAT enrichment experiments, we found proteins involved in the biosynthesis of pyochelin, a secondary siderophore involved in bacterial iron acquisition, were up-regulated in the regrowth phase. We further explored whether the up-regulation was a result of the modulation of HigB-HigA toxin-antitoxin system.
In Chapter 4, we describe our work for understanding molecular redundancy. The chapter follows up on our observation of up-regulation of pyochelin-related proteins during persister regrowth. We discuss the hypothesis that pyochelin confers a growth advantage in persister cells subject to carbon-limited conditions. In addition, we discuss the potential role of Fur, a ferric uptake regulator, in bacterial persistence.</p
Rare Higgs Processes at CMS and Precision Timing Detector Studies for HL-LHC CMS Upgrade
Get PDFThis thesis describes the search for two rare Higgs processes. The first analysis describes the CMS Run 2 search for H → µµ decays, with 137.3 fb-1 of data at √s = 13 TeV. The analysis targeted four different Higgs production modes: the gluon fusion (ggH), the vector boson fusion (VBF), the Higgs-strahlung process (VH), and the production in association with a pair of top quarks (ttH). Each category used a dedicated machine learning based classifier to separate the signal from the background processes. A combined fit from all these categories saw a slight excess in the data corresponding to 3.0 standard deviations at MH = 125.38 GeV, and gave the first evidence for the Higgs boson decay to second-generation fermions. The best-fit signal strength and the corresponding 68% CL interval was found to be µ^ =1.19 +0.41-0.39 (stat) +0.17-0.16 (syst) at MH = 125.38 GeV.
The second analysis describes the CMS Run 2 search for HH → bb̅bb̅ with highly boosted Higgs bosons. This analysis used a dedicated jet identification algorithm based on graph neural networks (ParticleNet) to identify boosted H → bb jets. This search targeted the gluon fusion and the vector boson fusion HH production modes, and put constraints on the allowed values of the various Higgs couplings as: κλ ∈ [-9.9,16.9] when κν = 1, κ2ν= 1; κν ∈ [-1.17,-0.79] ∪ [0.81,1.18] when κλ = 1, κ2ν = 1; κ2ν ∈ [0.62,1.41] when κλ = 1, κν = 1. A scenario with κ2ν = 0 was excluded with a significance of 6.3 standard deviations for the first time, when other H couplings are fixed to their SM values. The combined observed (expected) 95% upper limit on the HH production cross section was found to be 9.9 (5.1) x SM.
Finally, this thesis also discusses the planned MIP Timing Detector (MTD) upgrade for CMS at the HL-LHC. The MTD will be a time-of-flight (TOF) detector, designed to provide a precision timing information for charged particles using SiPMs + LYSO scintillating crystals, with a time resolution of ~30 ps. This thesis describes several R&D tests that have been performed for characterizing the sensor properties (time resolution, light yield, etc.) and optimizing the sensor design geometry. This thesis also contains a description of mock test setups for cooling the sensors, since it is known to be an effective way of mitigating the increased dark current rates in the sensors due to radiation damage. </p
Development of Numerical Models to Advance the Understanding of Air Quality in Los Angeles
Get PDFAtmospheric pollutants such as particulate matter (PM) and ozone (O₃) are harmful to human health and intensify climate change. Secondary organic aerosol (SOA) is a main component of PM and is formed via atmospheric oxidation reactions of thousands of gas- and aerosol-phase precursors. Regional-scale chemical transport models predict the formation of these pollutants by representing natural and human emissions of hundreds of species and their subsequent chemical and physical processing in the atmosphere. These models are useful in the absence of detailed measurements and allow researchers to investigate the impact of changing emissions and weather. Los Angeles has unique meteorology and anthropogenic emissions which lead to dangerous pollution events and make this region an important area to study SOA, PM, and O₃ formation. As vehicles have become cleaner and their emissions have declined, other sources of emissions have become increasingly important. One important category of emissions is volatile chemical products (VCPs), which are consumer and industrial products that have high volatile organic compound (VOC) emissions that have not been well-constrained or studied in relation to their SOA and O₃ formation potential. In this dissertation, I use the Community Multiscale Air Quality (CMAQ) model to represent the air quality of the Los Angeles Basin. First, a new chemical mechanism is developed to represent the formation of SOA from VCPs, implemented in the CMAQ model to simulate 2010 California, and the impact of VCPs on atmospheric pollutants is quantified. Next, we created contemporary inputs to CMAQ by simulating the meteorology, emissions, and land surface of the Los Angeles Basin in 2020. Lastly, the new inputs and chemistry are applied to CMAQ to understand current air quality issues in Los Angeles. We quantify the impact of VCPs on SOA, PM, O₃, and other pollutants in both 2010 and 2020. The apportionment of other emission sources and the impact of the COVID-19 pandemic are investigated, and pollutant concentrations are compared to measurements made throughout the Basin and specifically in Pasadena. This works demonstrates the importance of intentional policies to mitigate harmful air pollution events. Limiting NOₓ emissions is not sufficient to limit the formation of ozone and PM, and there must be a simultaneous reduction of VOC emissions
Mechanism and Function of Nascent Protein Modification in Bacteria
Get PDFNewly synthesized proteins undergo multiple modifications to ensure proper biogenesis and acquire their functions. N-terminal methionine excision (NME), mediated by the sequential actions of peptide deformylase (PDF) and methionine aminopeptidase (MAP), is an essential and the most prevalent N-terminal protein modification in the bacterial proteome. Despite the extensive studies on enzymatic catalysis, how NME impacts various cellular functions and how the enzymes achieve timing and selectivity under complex cellular conditions have been long-standing puzzles.
In this work, we use a combination of biochemical analyses, computational modeling, and in vivo measurements to investigate the molecular mechanisms and physiological functions of cotranslational NME reactions. We show that the interactions between the ribosome, the nascent chain, the NME enzymes, and other ribosome-associated protein biogenesis factors dramatically remodel the kinetics and specificity of NME reactions under physiological conditions. In addition, we apply time-resolved, system-wide analyses on the translatome and steady-state proteome to study how the inhibition of PDF influences diverse cellular pathways in bacteria. The results unveil the impact of NME on the biogenesis of nascent proteins and highlight the role of the membrane in coupling the biochemical activities of NME enzymes to cellular physiology.</p
Production and Characterization of Ytterbium Monohydroxide (YbOH) for Next-Generation Parity and Time-Reversal Violating Physics Searches
Get PDFNew sources of parity (P) and time-reversal (T) violating physics are motivated by several unanswered questions in fundamental physics, including the observed imbalance between matter and anti-matter in the universe. P,T-violating effects can induce permanent electric dipole moments (EDMs) in atoms and molecules, allowing them to act as sensitive probes of new physics. The linear, triatomic molecule ytterbium monohydroxide (YbOH) has emerged as a promising candidate for next-generation molecular EDM searches, because it possesses both an electronic structure amenable to optical cycling and parity doublets in the bending mode. These features enable laser cooling, highly polarizable science states, and internal comagnetometry which promises an order-of-magnitude improvement to current EDM sensitivities. Additionally, different isotoplogues of YbOH offer sensitivity to different sources of P,T-violating physics: leptonic sources via a measurement of the electron’s EDM in 174YbOH and hadronic sources via a measurement of the nuclear magnetic quadrupole moment (NMQM) of the 173Yb nucleus in 173YbOH. In this dissertation, I describe the design, construction, and optimization of a YbOH cryogenic buffer gas beam (CBGB) source, including the implementation of laser-enhanced chemical reactions for increased molecular production. Direct and frequency modulated (FM) absorption spectroscopy and laser-induced fluorescence measurements (LIF) were implemented in the CBGB source, and LIF and separated field pump/probe microwave optical double resonance spectroscopy was conducted in a supersonic molecular beam source. Additionally, laser-enhanced chemical reactions were utilized to develop a novel spectroscopic technique critical to the observation of the spectrum of the odd isotopologues. FM absorption spectroscopy in the CBGB source allowed the observation of the previously unobserved, weak Ã2Π1/2(1,0,0)-X̃2Σ+(3,0,0), [17.68], and [17.64] vibronic bands. The X̃2Σ+(0,0,0) ground state has been characterized at high precision and the Ã2Π1/2(1,0,0)-X̃2Σ+(3,0,0) band of 174YbOH and the Ã2Π1/2(0,0,0)-X̃2Σ+(0,0,0) band of the odd 171,173YbOH isotopologues have been characterized for the first time. This work provides much of the spectroscopic knowledge needed to implement next-generation P,T-violating physics searches in YbOH
The Chinese Room Argument
No full textIntroduction: The question of what constitutes understanding is often discussed without considering the more fundamental question of what the human mind is. Searle’s Chinese Room Argument attempts to challenge those who believe in strong AI and functionalism by proposing an example that meets their requirements for understanding yet intuitively seems to lack understanding. In the process, Searle makes claims about what AI functionalists believe about the nature of the human mind as well as how he differs from them. This essay will discuss the views of Searle and his detractors, and attempt to extend those views and dissect how they interact with each other
Resilience of a Precise Motor Behavior
Get PDFMotor memory retention is an essential part of survival and reproduction of most species. However, these behaviors are variable and hard to measure. The zebra finch provides a great model organism to study motor behavior on a fine scale and ask fundamentally important questions. Zebra finch males learn their song from their father and once learnt this song remains unchanged for the remainder of the animals’ life. This highly stereotypic and precise motor function engages a handful of motor nuclei organized in a spatially spread out manner that allows for precise targeting of each key circuit participant for the production of the behavior. In my studies, I focus on better understanding the role of excitatory and inhibitory neurons in the pre-motor nucleus of the song production system. The goal was to perturb the precision of behavioral execution by collapsing the neuronal circuit responsible for sequential activity. Then, to study if the behavior could re-establish in an adult less plastic state of neuronal organization. After I have shown that motor function recovers to produce the same song post disruption, I investigated the large and small scale changes in neuronal activity and transcriptomics accompanying this degradation and recovery trajectory. I have learned that loss of inhibition leads to hyperactivation which eventually leads to a circuit level homeostatic compensation to shut down the pathological activity level. In addition, the upregulation of MHC1 receptors and microglia points to a homeostatic mechanism for synaptic reorganization and re-establishment. Now that we have the means to execute precise cell-type specific manipulations that are reversible and that we understand the underlying phenomenology of perturbation and recovery, we can ask many questions about the architecture of a highly resilient motor pathway. This could shine light on specific electrophysiological and molecular candidates to study for brain damage repair and neurodegenerative research