80264 research outputs found
Sort by
MetaHeart: Spoofing Vibrational Biometrics via Dynamic Metasurfaces
No full textPrivacy-invading biometrics monitoring is becoming a prominent security threat as modern sensing systems move to higher operating frequencies (mmWave, sub-THz), increasing sensing resolution and accuracy. As such, developing systems that can protect or obfuscate biometrics from adversarial intrusion becomes pivotal to preserving user privacy. In this work, we develop and implement MetaHeart, a real-time biometrics misinformation system based on reflective, programmable metasurfaces and dynamic phase-front manipulation of radar inferences. MetaHeart’s key goal is to prevent the leakage of a legitimate user’s heartbeat biometrics by spoofing fake heartbeat signals at a malicious, radar-equipped, heart rate sensing intruder. We experimentally demonstrate MetaHeart’s ability to fake Alice’s presence when she is not there and to fool Trudy’s inferences even when Alice is present, achieving an
overall accuracy above 98%. Finally, we conduct a robustness analysis to determine MetaHeart’s required spatial
placement within the intruder’s monitoring area that would allow for effective spoofing
Using systems and synthetic biology to understand the transcriptional responses of Variovorax beijingensis to changes in osmotic and matric potential
No full textSoil is a complex and dynamic environment in which microbes continuously adapt to fluctuating conditions. Soil microbial activities are instrumental to Earth carbon and nitrogen cycles, and they contribute significantly to soil health and plant fitness. With climate change, it is critical to understand how soil microbes respond to shifts in water availability as the extreme hydrological events intensify. Microbe-available water can be measured with soil water potential - the force with which soil particles hold on to the water. This force is influenced by both osmotic potential, driven by solute concentration, and matric potential, controlled by soil texture. Currently, it is unclear if these processes have similar effects on soil microbial behaviors.
In this thesis, I present systems biology experiments that examine transcriptional responses of a soil microbe (Variovorax beijingensis) to shifts in osmotic and matric potential. By analyzing global transcription in liquid medium and in artificial soils that induce a similar range of water potential shifts, I find that V. beijingensis exhibits substantial overlap in its response to these two types of stress. However, the responses are not identical, matric potential shifts result in a broader set of differentially expressed genes. I also describe synthetic biology experiments to engineer soil microbes to report on their promoter activities in soil within the context of a model soil community. I show that V. beijingensis can be programmed to report on a putative hyperosmotic response promoter by producing methyl halide, an indicator gas. This gas can be detected in the headspace of soil habitats using analytical instrumentation or a soil-derived Methylorubrum that is programmed to produce a visual output in response to methyl halides.
My systems biology results reveal a set of differentially expressed genes that arise from increases in water potential, independent of the mechanism by which the water potential is changed, illustrating a shared response to both types of pressure changes. My synthetic biology results complement these efforts by yielding tools that provide a foundation for future studies investigating responses of model soil communities to dynamic hydration conditions, such as those driven by drought and flood events
Shortest-Path Solutions for Connecting Multiple Terminals via Deep Learning
No full textConnecting multiple terminals with the shortest possible overall path in an obstacle-strewn indoor environment is a classic path planning problem. Standard solutions such as A*, Anytime Repairing A* (ARA*), and Rapidly-exploring Random Tree* (RRT*) have shorter runtimes when input environment maps are smaller and exclude extraneous potential paths. Existing obstacle-avoiding rectilinear Steiner minimum tree (OARMST)-solving algorithms that could be used to simplify indoor environments are, however, approximate algorithms that can produce incorrect shortest-path solutions. This work proposes a deep learning-based front-end method called MazeNet that successfully constrains indoor search spaces, thereby improving runtimes without adding the expense of longer paths. MazeNet employs a recurrent convolutional neural network (RCNN) trained to solve (synthetic) grid mazes with the shortest feasible path connecting all specified terminals. The resulting predicted tree imposes structural constraints on the floorplan, restricting the search to subregions that contain the optimal shortest path solution. This method is compatible with a wide range of classical planners and can be deployed as a preprocessing module to accelerate real-time navigation. In more complex real environments, such maze solutions can be used to deliver simplified floorplans, enabling shorter runtimes for algorithms such as A*. Experimental results on both synthetic mazes and discretized real-world floorplans demonstrate that MazeNet consistently delivers the ground-truth shortest path and outperforms approximation-based filters in solution accuracy
Demography as Destiny and The Latino Shift: Mapping, Understanding, and Tracking Shifting Campaign Contributions
No full textThis thesis examines whether the widely discussed rightward drift of Latino voters in the 2024 election has taken root in campaign contributions—a more costly and enduring form of political engagement. Drawing on over 78 million FEC contribution records from the 2020 and 2024 election cycles, I map partisan giving across “Border Texas” counties and compare Hispanic and non-Hispanic donation patterns. While Latino Republican donors in the region gave more on average in 2024 than in 2020, their overall share of Republican contributors declined significantly, suggesting a deepening of commitment rather than a broadening of support. Spatial analysis reveals that Hispanic giving clustered geographically, whereas non-Hispanic giving did not, underscoring the role of local context in partisan realignment. These findings complicate the emerging narrative of a wholesale Latino shift to the Republican Party, showing instead a political identity in flux, with donations lagging voting behavior and signaling a realignment still unsettled
Data for Lattice-induced spin dynamics in Dirac magnet CoTiO3
No full textRaw data for the figure plots in the journal article doi: 10.1063/5.028288
Multifunctional Textile-Based Wearable Devices Enabled by Thermal Design
No full textWearable assistive devices and robots are increasingly composed of compliant materials to enhance user comfort, with 2D sheet-based versions of these materials offering ease of integration into clothing architectures. While sheets of materials offer enhanced compliance and integration, a need remains for a customizable and cost-effective design approach to develop functional sheet-based systems composed of off-the-shelf materials. To bridge this gap, my doctoral thesis focuses on the thermal design of wearable textile-based devices realized through the following three objectives: (1) Designing textile-based materials for rapid in situ thermal self-decontamination using Joule heating. I developed a textile-based material that achieves temperatures (≥100 °C) necessary for rapid thermal inactivation of viruses in short durations (≤5 s) via Joule heating. I investigated the coupling of the geometry of the material to its thermal performance, and corresponding ability to inactivate viruses, all without harming the user nor requiring them to doff the material during the decontamination process. (2) Integrating temperature control in situ to enable thermal self-sensing within a wearable heating material for controlled thermoregulation. The relationship between the electrical resistance of the heating material and its temperature allows the material to be used as both a thermal actuator and thermal sensor. Integrating the temperature-dependent changes of resistance with closed-loop control realizes a self-modulating multifunctional heating material for targeted thermal performance. (3) Powering soft fluidic assistive actuators with waste body heat. Long-lasting modes of portable power are necessary for increased adoption and utility of fluidic assistive actuators. Waste body heat, a readily available source of power for wearables, is directly converted to and stored as pneumatic energy using a textile-based system charged with a low-boiling-point working fluid. Overall, my advancements through these three
objectives have opened a pathway to develop sheet-based wearable devices via a thermal design approach with potential to improve the standard of living for people with thermoregulatory and mobility-related limitations, as well as the capabilities of everyday wearable technology
Evaluating Runoff Response to Nature-Based Solutions Under Varying Development Scenarios in Upper Cypress Creek near Houston, Texas
No full textAs climate change is driving an increase in the intensity and frequency of rain events, urban planners have been examining nature-based solutions (NBS) as possible tools to address the growing flood risk. NBS such as rain gardens, wetlands, riparian setbacks, and natural ecosystem restoration all attenuate flood waves by storing additional water in soil or soil-mimicking technology. Hydrologic processes like infiltration, re-infiltration, and subsurface flow are critical to how NBS impact runoff response and thus must be accurately represented in models. These processes are not always represented in common flood inundation models. In this study, we investigate the impact of four development scenarios within areas protected for restoration of native prairie grasslands, and areas in potential hotspots for future development in the region. The goal is to assess the impact the protected natural areas have on the runoff response of Upper Cypress Creek. We also compare how two different modeling frameworks represent the impact of NBS on the runoff response. To do this a coupled hydrologic-hydraulic framework, CREST-iMAP, is benchmarked against a FEMA-approved hydrodynamic modeling software, HEC-RAS. This study found that the development of protected grasslands resulted in increased flows, decreased lag times, as well as significant attenuation of surface runoff from upstream development. Indicating that these natural ecosystems store and detain a significant amount of water and can be an integral component of a flood mitigation strategy. Compared to HEC-RAS, CREST-iMAP predicted significantly less overland flow, as well as noticeable differences in flow reduction between the development scenarios. This suggests that flood inundation models that more accurately represent hydrologic processes and soil moisture dynamics could better evaluate the effectiveness of nature-based solutions
Program Evaluation Methods with Network Interference
Get PDFThis dissertation examines the identification and estimation of causal effects of programs or policies in settings where individuals interact, violating the classical Stable Unit Treatment Value Assumption (SUTVA). Such interactions create spillover effects that standard causal inference methods may overlook.
The first chapter extends the local average treatment effect framework with imperfect compliance to settings where two individuals interact. I propose a weak monotonicity concept that generalizes assumptions on treatment take-up, such as one-sided noncompliance, commonly used to identify causal effects. I show that local average treatment effects can be identified under this weak monotonicity, given an additional exclusion restriction for endogenous treatment. I propose an estimator and apply it to experimental data from a Kenyan savings account study, demonstrating its effectiveness in addressing imperfect compliance and interference with more accurate causal estimates.
The second chapter develops a causal inference framework under interference, allowing network changes due to treatment. Unlike studies assuming a fixed network, I consider structural changes in the network caused by treatment. I decompose causal effects into two parts: one when the network remains unchanged (treatment effects) and another from network changes (network effects).
I propose that this decomposition can be identified when the potential outcome is determined by a linear response function, under two experimental settings: a randomized experiment and a quasi-experiment.
I propose an estimator and its performance is verified through simulations and the experimental data from Nepal’s savings account program. In the empirical application, I find a small total effect composed of two significant effects in opposite directions.
The third chapter relaxes the data requirements of the second chapter. Instead of full network information, I assume researchers observe exposures that are some summary statistics like the number of treated friends. Treatment may alter the network, affecting exposures, which in turn influence outcomes. I interpret exposure as a mediator and use a causal mediation framework to decompose the total effect into direct and mediated effects. Like in the second chapter, the mediated effect represents the network effect. Applying this method to coeducational high schools' impact on academic performance, I find differential effects by gender
1.2 Risk-benefit analysis worksheet for biotechnology beyond conventional containment
No full textThis entreaty was created as part of The Spirit of Asilomar and the Future of Biotechnology summit (February 23-26, 2025) in Pacific Grove, CA.This report presents a structured risk-benefit analysis framework developed through discussions at the Spirit of Asilomar summit under the Biotechnologies Beyond Conventional Containment (BBCC) theme. As biotechnologies increasingly move beyond traditional laboratory and industrial confinement into open environmental applications, novel risks, ethical considerations, and societal implications arise. Recognizing the limitations of defining universal "safe design guidelines”, this report focuses instead on creating a practical, context-sensitive worksheet to guide scientists in early-stage project development. The worksheet prompts users to evaluate potential benefits—environmental, health, economic, and societal—against biological, ecological, health, regulatory, and security risks, while acknowledging the inevitability of residual risks. It encourages interdisciplinary consultation, highlighting the importance of engaging with experts and stakeholders to ensure responsible innovation. Serving as a preliminary self-assessment tool, the worksheet is designed to complement, not replace, formal regulatory processes. It aims to foster more thoughtful and transparent biotechnology design decisions and to evolve over time through broader community input
Towards the Controlled Synthesis and Industrial Applications of Two-Dimensional (2D) Materials Guided by Machine Learning
No full textSince the first exfoliation of graphene from graphite in 2004, atomically thin two-dimensional (2D) materials have gained significant attention due to their unique properties that emerge during the transition from bulk to monolayer form. These characteristics have enabled a broad range of applications spanning nanoelectronics, optoelectronics, and energy systems. While substantial progress has been made in the discovery and synthesis of 2D materials, challenges remain in achieving controlled growth and scalable, cost-effective production.
Recent advances in integrating machine learning (ML) into practical engineering processes have accelerated its adoption in 2D materials research by reducing the labor required for large-scale data analysis. This thesis addresses the pressing challenges in 2D material development by establishing a data-driven framework to investigate both the top-down exfoliation and bottom-up synthesis mechanisms, while also exploring their potential for industrial applications.
A customized miniature CVD platform was developed to facilitate real-time optical monitoring of MoS₂ monolayer growth. Through image processing techniques, real-time growth footage was digitized, enabling the extraction of key morphological parameters such as nucleation density, crystal coverage, and growth rate. Machine learning algorithms were employed to correlate these parameters with process conditions, enabling predictive modeling and CVD optimization. This closed-loop system lays the foundation for future autonomous material synthesis platforms.
In parallel, a polymer-assisted dry ball-milling method was introduced for scalable exfoliation of hexagonal boron nitride (hBN). Using ML-based feature selection, critical polymer properties responsible for high exfoliation efficiency were identified. The method was successfully extended to exfoliate various layered materials, supporting its potential for large-scale manufacturing. Additionally, atomic-layer structured photovoltaics (ALSPs) incorporating 2D heterostructures were fabricated, demonstrating strong performance, stability, and flexibility, with promising applications in self-powered electronics.
In general, these studies establish a robust, data-driven strategy for controlled synthesis and industrial implementation of 2D materials