137091 research outputs found
Sort by
Interdisciplinary Investigations into Sensory Neuron Mechanisms and Modulations in Health and Disease
No full textThe somatosensory system is a critical component to maintaining homeostasis and is often one of the first responders to changes in external and internal stimuli, quickly triggering the body to mitigate any threats. However, despite continuous and significant advancements in our understanding of the somatosensory system’s role in various facets of health and disease from skin disease to non-invasive technologies for sensory neuron manipulation, many of the complex mechanisms still remain widely unknown, which hinders novel and effective treatments that may reduce economic and health burdens. In this dissertation, I attempt to answer some of these critical questions that remain unanswered by (1) investigating toxins released by S. aureus, which may be implicated in atopic dermatitis (AD), (2) conducting a function analysis of Mas-related G protein-coupled receptor (Mrgpr) C11+ neurons within the vagal ganglia (VG) that innervate and contribute to the function of the airway, and (3) revealing ultrasound parameters that may be critical to FUS-induced neuromodulation. These findings reveal new avenues for research in a wide range of areas involving the somatosensory system, may provide novel tools for studying the role of the somatosensory system in facets of health and disease, and offer crucial insights for future therapeutic efforts.Ph.D.Biolog
How do Practitioners Define and Design for Privacy When Developing Consumer AI Technologies?
No full textAs artificial intelligence (AI) is swiftly advancing many facets of technology, it is more pertinent than ever to consider how these systems are designed for end user privacy. Prior research reveals that there are areas for AI practitioners to improve privacy considerations when designing AI systems, but lack clarity in how these changes could be implemented. We looked to reveal exactly where the gap between research and practice is for AI practitioners and privacy. By conducting semi-structured interviews, we revealed that this gap often stems from a lack of a centralized privacy definition. While practitioners may have felt personally motivated to uphold privacy, the lack of privacy education and compliance-centered rigidity made it difficult to do so. Moving forward, we believe that informed, up-to-date policy could be a good path to ensure privacy is upheld in all AI products.UndergraduateComputer Scienc
Video Conferencing using Edge Computing Infrastructure in the Atlanta Westside Communities
No full textThe COVID-19 pandemic has introduced a more hybridized learning and working environment in our lives, with people taking advantage of virtual workspaces, tools and resources, and cloud infrastructure to host meetings online. However, the digital divide and corresponding achievement gap in education are further exacerbated as those without stable access to community networks are left behind. My goal, with the Embedded Pervasive Lab, was to research video conferencing platforms that can utilize the power of edge computing and to better understand what the threshold of these platforms looks like, with the running hypothesis that the software should be able to sustain a medium-sized conference call. Load testing was performed for Ion video conferencing software, and the data suggested that while this tool can support smaller conferences with fewer clients, it encounters high failure rates and rapidly decreasing bandwidths for higher client simulations. Ultimately, the data cannot support the hypothesis, outlining a use case for this software. This study opens future research in testing other software, adding functionality and better controls on Ion conferencing, and looking at other metric testing criteria.UndergraduateComputer Scienc
Applications of New Materials Systems and Machine Learning Techniques for compensation and Calibration of MEMS
No full textThis PhD dissertation explores significant advancements in the thermal stability and calibration accuracy of MEMS (Micro-Electro-Mechanical Systems) devices, particularly focusing on the application of high-performance 4H-SiC (silicon carbide) and AlN (aluminum nitride) in MEMS resonators and gyroscopes. Moreover, the research provides a high-level exploration of the critical advancements in MEMS technology, focusing on enhancing thermal stability and calibration accuracy through innovative material applications and cutting-edge ML and DL techniques. The work presented in this dissertation collectively contributes to the development of highly stable and accurate MEMS resonators and gyroscopes, paving the way for their broader application in various high-stakes and everyday environments. These innovations ensure the continued evolution and reliability of MEMS technology, highlighting its importance in modern technological applications.Ph.D.Electrical and Computer Engineerin
A Physical Design Framework For Creating Fine-Grained Multi-Tier SRAM Arrays
No full textThis study presents a novel approach to address the challenges of scaling limitations and high read/write latencies inherent in conventional 2D SRAM designs through the development of a novel 15nm FinFET-based standalone 3D SRAM Array. Utilizing Monolithic Intertier Vias, our two-tiered implementation effectively mitigates these limitations along both X and Y axes. We introduce two distinct design methodologies for 3D SRAM arrays - Wordline and Bitline folding. Through post-layout simulations conducted on arrays with capacities of 2kB (256WLx64BL) and 8kB (512WLx128BL), our 3D designs exhibit superior performance metrics. Notably, the 3D Wordline-Folded design achieves a remarkable 57.54% average reduction in footprint compared to the 2D baseline across both array sizes. Furthermore, the 3D Bitline-Folded configuration demonstrates consistent superiority in speed, with an average read latency improvement of 17.16% and a notable 54.2% enhancement in write latency. Conversely, the 3D Wordline-Folded array emerges as the most energy-efficient option, boasting an average reduction of 15.3% in read energy and over 21% in write energy compared to the 2D baseline.M.S.Electrical and Computer Engineerin
Measuring Dynamic Team Reorganization and Interdependency in Response to Uncertainty
No full textTeams working in complex and dynamic environments often encounter uncertainty, and in many cases must adapt to overcome it to maintain team effectiveness. One fundamental component of adaptation is reorganization. Specifically, it hypothesized that teams adapt by changing their coordination patterns. One way teams can reorganize is by increasing or decreasing team member role interdependency, by tightening or loosening coordination constraints depending on the situation. Dynamic interdependency is hypothesized to be necessary for proper team coordination and functioning, especially during times of high uncertainty to avoid potentially dangerous and disastrous consequences. This research focused on the development of measures of team-level reorganization in response to uncertainty and dynamic interdependency in teams. It was hypothesized that teams who display greater reorganization and properly modulated levels of dynamic interdependency during times of increased uncertainty will maintain high team performance during uncertain situations. To investigate different types of uncertainty, this dissertation utilized sources of uncertainty from a taxonomy of team uncertainty based in the team science literature (Grimm, 2022). This taxonomy informed the design and implementation of different types of uncertainty in a controlled experiment.
Two studies are included in this dissertation. The first study was a pilot study (Study 1) and was an analysis of archival data of teams working in an air battle management scenario (D. A. Grimm et al., 2024). The purpose of Study 1 was to formulate and initially validate measures of team reorganization and dynamic interdependency in response to uncertainty. Both measures were correlated with team performance, establishing their validity for detecting dynamic changes in team coordination in response to uncertainty. Results from Study 1 provided support for the measures of reorganization and dynamic interdependency across technological and communication-based components of the task, as they were positively related to performance. These findings helped inform the hypotheses for Study 2.
In the second study, a controlled laboratory experiment was carried out that involved 10 teams of 3 people (one trained confederate, two human participants) completing one seven-hour experimental session. Teams completed the experiment in a simulated remotely piloted aircraft (RPA) system simulation testbed and encountered various types of uncertainty-inducing perturbations. The experiment built on Study 1 by implementing perturbations specifically informed by the team uncertainty taxonomy. The goal of Study 2 was to examine how different types of uncertainty are related to different patterns of team reorganization and interdependency across different layers of system coordination (verbal communication, RPA behavior). The uncertainty perturbations were used to test the hypothesized relationships between team reorganization and dynamic interdependency in response to uncertainty by predicting team effectiveness (outcome and process performance). It was found that the communication layer (Gorman et al., 2019) primarily reflected different patterns of reorganization and interdependency, supporting the study hypotheses. Increased interdependency was also shown to occur in the technological layer in response to uncertainty, though in ways not in exact accordance with the study hypotheses. Finally, it was found that rapid and timely communications, as well as specific descriptions of technological or teammate-related uncertainties, significantly predicted the team-level ability to overcome perturbations.
The implications of these findings are discussed in terms of real-time reorganization and interdependency measurement for teaming operations in complex, degraded, and uncertain task environments. The results of this research have implications for teams working in task environments where uncertainty is inevitable and team competencies, such as reorganization and interdependency, are vital. Areas of application may include military, healthcare, energy, and transportation.Ph.D.Psycholog
Probing the Structure-Property Relationship of Porous Sorbents for Atmospheric Water Adsorption
No full textThis dissertation investigates the structure-property relationship of porous sorbents and their water adsorption behavior to develop materials for atmospheric water harvesting. Currently, the global water crisis is the biggest threat to mankind, with over half of the world’s population experiencing water scarcity at some time in the year. This issue is exacerbated by climate change, rapid population growth, and inadequate water infrastructure, and it is only expected to become more dire without proper scientific intervention. Common solutions, such as desalination, are often difficult to implement in remote or inland regions, arid regions, or areas where infrastructure is not abundant. Because of this, the development of adsorption based atmospheric water harvesting (AWH) technologies is becoming increasingly appealing.
Adsorption-based AWH technologies are explored in depth in Chapter 1. These technologies are attractive due to their ability to generate water from all environmental conditions, including arid air, and their continual renewability. An ideal water harvesting material must have high water capacity, have fast kinetics and mass transfer, be cyclically stable, and be able to be regenerated easily. To that end, this dissertation focuses on two classes of porous sorbents, hierarchical silicas and metal-organic frameworks (MOFs), for water adsorption applications, which are discussed in Chapter 2.
One of the biggest limitations of porous sorbents is their difficulty adsorbing significant quantities of water vapor from arid air. Materials that can adsorb water vapor from arid air typically possess small pore volumes which limit their maximum water capacity. On the other hand, materials that have large pore volumes and can capture large quantities of water require exposure to humid air. To that end, Chapter 3 focuses on developing a hierarchical silica-salt composite with record-setting water capacities across the full relative humidity regime. Through the incorporation of a hygroscopic salt into the pores of the silica matrix, there is a synergistic effect between the pore structure and hydrophilicity, harnessing the benefits of both the salt and silica sorbent. Similarly, this composite also exhibits cyclical stability, which suggests that the salt is stable within the pores of the silica and can continually be exposed to water adsorption and desorption.
Next, to understand how specific structural properties can be synthetically introduced to tune the water adsorption behavior, defects were engineered into MOF structures via soft templating. In Chapter 4, robust, water stable MOFs are synthesized in the presence of CTAB surfactants to introduce structural defects to modify pore size and volume. Large macropores and interparticle voids were generated with smaller MOF crystals, resulting in changes to the water adsorption behavior. Lastly, in Chapter 5, a MOF was post-synthetically modified through cation exchange allowing for the tailoring of the water adsorption behavior in a singular framework, resulting in increased water capacity, better performance in arid conditions, and increased hydrolytic stability.
In summary, by exploring three distinct methods of material engineering, namely hygroscopic salt impregnation, synthetic defect engineering, and post-synthetic cation exchange, this dissertation explores the intricacies of specific material properties on water adsorption behavior to develop a better understanding of how adsorption-based atmospheric water harvesting can be used to mitigate the global water crisis.Ph.D.Chemical and Biomolecular Engineerin
Conditioning Multi-Agent Policies on Capabilities Enables Zero-Shot Generalization to New Robot Teams
No full textIn this work, we use MARL to solve a decentralized, partially-observable coordination problem for a team of heterogeneous robots, with a policy that can zero-shot generalize to different robot types and team compositions. To do so, we leverage the centralized training, decentralized execution (CTDE) paradigm to create a decentralized policy which can be copied to any number of robots, and condition a graph neural network (GNN) based policy on the capabilities of each robot to allow a single learned policy to zero-shot generalize to unseen robots and team compositions. We call this zero-shot generalizability adaptive teaming. Like prior work, our GNN facilitates learned communication of partial observations. Unlike prior work, we also demonstrate that communicating capability information among agents is vital for effective coordination in a heterogeneous multi-robot team. We find the constraint of needing to know the capabilities of each robot fairly realistic for real robotics use, as often real robots come with a specification sheet of known properties, such as maximum sensing range or top speed. We evaluate our policy's adaptive teaming ability on a cooperative robotics task we name Heterogeneous Sensor Network (HSN), a heterogeneous-robot variant of the common sensor coverage task. Our results demonstrate that our policy architecture, with awareness and communication of individual robot capabilities, outperforms existing agent-ID methods on adaptive teaming. Additionally, we provide results indicating that, allowing robots to communicate their own capabilities improves performance even without considering generalization as a necessary component. Finally, we share results of our policy coordinating real robots in the Robotarium, showing that our technique can be applied to real robots.UndergraduateComputer Scienc
Capability-Aware Shared Hypernetworks for Heterogeneous Multi-Agent Coordination
No full textCooperative heterogeneous multi-agent tasks require agents to behave in a flexible and complementary manner that best leverages their diverse capabilities. Learning-based approaches to this challenge span a spectrum between two endpoints: i) shared-parameter methods, which assign an ID to each agent to encode diverse behaviors within a single architecture for sample-efficiency, but are limited in their ability to learn diverse behaviors; ii) independent methods, which learn a separate policy for each agent, enabling greater diversity at the cost of sample- and parameter-efficiency. Prior work on learning for heterogeneous multi-agent teams has already explored the middle ground of this spectrum by learning shared-parameter or independent policies for classes of agents, allowing for a compromise between diversity and efficiency. However, these approaches still do not reason over the impact of agent capabilities on behavior, and thus cannot generalize to unseen agents or team compositions.
In this work, we aim to enable flexible and heterogeneous coordination without sacrificing diversity, sample efficiency or generalization to unseen agents and teams. First, inspired by work from trait-based heterogeneous task allocation, we explore how capability-awareness enables generalization to unseen agents and teams. We thoroughly evaluate our GNN-based capability-aware policy architecture, showing that it can more effectively generalize than existing work.
Then, inspired by recent work in transfer learning and meta-RL, we propose Capability-Aware Shared Hypernetworks (CASH), a new soft weight sharing architecture for heterogeneous coordination that use hypernetworks to explicitly reason about continuous agent capabilities in addition to local observations. Intuitively, CASH allows the team to learn shared decision making strategies (captured by a shared encoder) that are readily adapted according to the team’s individual and collective capabilities (by a shared hypernetwork). Our design is agnostic to the underlying learning paradigm. We conducted detailed experiments across two heterogeneous coordination tasks and three standard learning paradigms (imitation learning, value-based and policy-gradient reinforcement learning). Results reveal that CASH generates appropriately diverse behaviors that consistently outperform baseline architectures in terms of task performance and sample efficiency during both training and zero-shot generalization. Notably, CASH provides these improvements with only 20% to 40% of the learnable parameters used by baselines.M.S.Computer Scienc
Living History Program
No full textThis series consists of oral histories completed by the Georgia Tech Alumni Association's Living History Program under the direction of Marilyn Somers