Doctor of Philosophy in Physics
School of Physics Dissertation Defense
Jiahao Chen
Dr. Dragomir Davidovic, School of Physics, Georgia Institute of Technology (Advisor)
Numerical Approach to the Time-Convolutionless (TCL) Master Equation
Date: Tuesday, November 10, 2026
Time: 8:30 a.m. EST
Location: Howey N201/202
Virtual: Join the Microsoft Teams meeting / Meeting ID: 273 524 787 663 512 4 / Passcode: ub6Fd2TB
Thesis Committee:
Dr. Brian Kennedy, School of Physics, Georgia Institute of Technology
Dr. Martin Mourigal, School of Physics, Georgia Institute of Technology
Dr. Xueda Wen, School of Physics, Georgia Institute of Technology
Dr. Qi Tang, School of Computational Science and Engineering, Georgia Institute of Technology
Abstract:
Open quantum systems are quantum systems that interact with their environment. The time-convolutionless (TCL) master equation provides a formally exact, time-local description of their dynamics. However, the computational cost of evaluating higher-order terms often limits practical applications to the second-order approximation. We develop systematic numerical methods that reduce the temporal complexity to O(N log N) for TCL4 and O(N log² N) for TCL6, where N is the number of time points. We extend the construction to higher orders and assess its accuracy using spin–boson benchmarks. We also investigate the convergence limits of the TCL expansion and explore resummation methods for extending accurate predictions beyond direct perturbative truncation.
]]>
Abstract:
Rare earth elements (REEs) are essential to advanced and clean energy technologies. Sustainable REE resource development requires both a fundamental understanding of the geochemical processes governing resource formation and advances in environmentally compatible extraction and recovery strategies. Highly weathered, clay-rich deposits represent important natural REE resources. However, the molecular-scale mechanisms by which solution chemistry, mineral surfaces, and naturally occurring ligands regulate REE behavior remain incompletely understood. Meanwhile, conventional REE extraction and recovery often involve high chemical and energy consumption and substantial waste generation. Bio-assisted approaches, including fungal bioleaching, offer promising alternatives that operate under milder conditions and may reduce chemical consumption and environmental impacts. However, the mechanistic links among cultivation conditions, microbial metabolism, and REE mobilization remain poorly understood.
This dissertation investigates the key geochemical controls on REE speciation, mobility, secondary phase formation during chemical weathering, and explores fungal bioleaching as a promising approach to REE extraction. Controlled laboratory experiments integrate solid-state nuclear magnetic resonance spectroscopy, synchrotron X-ray absorption spectroscopy, microscopy, aqueous-phase analyses, and extracellular metabolomics to connect molecular-scale reaction mechanisms controlling REE behavior at mineral-water interfaces and extraction performance. Using yttrium (Y) as a probe for heavy REEs, phosphate and citrate as representative ligands, kaolinite as a mineral substrate, this work identifies complementary mechanisms governing REE retention and release during weathering. Phosphate promotes Y immobilization at kaolinite-water interfaces through ternary surface complex and secondary Y-phosphate precipitation. In contrast, citrate can promote either Y retention or remobilization, depending on ligand concentration, solution pH, and reaction sequence. Reactant concentrations and solution pH further regulate the crystallinity of newly formed REE-bearing secondary phase. Extending this investigation to biologically mediated mobilization, experiments with Aspergillus niger and natural REE-bearing mineral sands suggest that initial pH act as a signaling cue shaping subsequent metabolic response and leaching trajectories. These findings suggest that cultivation conditions could be used to guide microbial metabolism and regulate the balance between REE dissolution and precipitation.
Together, this work advances the mechanistic understanding of REE speciation, mobility, redistribution, and secondary-phase formation in weathering systems and provides a basis for tuning bioleaching systems to achieve desired extraction outcomes.
Committee members: Dr. Yuanzhi Tang (Advisor), Dr. Martial Taillefert, Dr. Jennifer B. Glass, Dr. Christopher T. Reinhard, and Dr. Brian L. Phillips
Date & Time: Friday November 13th, 2:00 pm
Location: ES&T L1114
]]>
Abstract:
Reliable geological carbon storage requires estimates of subsurface CO₂ saturation and pressure despite incomplete observations and uncertain reservoir properties. This dissertation develops uncertainty-aware Digital Shadows and Digital Twins that combine physics-based simulation with amortized Bayesian inference for monitoring evolving reservoir states.
The proposed framework uses ensembles propagated through multiphase flow, rock physics, seismic modeling, and imaging to train conditional normalizing flows that generate posterior samples from time-lapse seismic and well observations. The dissertation examines three extensions of this framework: multimodal monitoring, robustness to uncertain rock physics, and context-aware inference. Rock-physics augmentation improves reconstruction under tested model mismatch, while a three-phase extension enables joint estimation of oil and CO₂ saturation. A sensitivity-aware Digital Twin treats rock-physics assumptions as explicit context, supporting assumption-specific inference and what-if analysis. Finally, a collaborative case study demonstrates how probabilistic Digital Shadow estimates can be supplied to a separately developed injection-control workflow. Together, these studies provide a simulation-based framework for uncertainty-aware subsurface monitoring and decision support.
Defense date: November 6, 2026
Defense time: 10:30 am - 01:30 pm EST
Defense location: ES&T L1114
Advisor (Chair):
Felix Herrmann, Professor, Georgia Institute of Technology
Committee members:
Andrew Newman, Professor, Georgia Institute of Technology
Zhigang Peng, Professor, Georgia Institute of Technology
Sheng Dai, Professor, Georgia Institute of Technology
Olav Moyner, Senior Research Scientist, SINTEF, Norway
]]>
Wenhao Qin
Advisor: Dr. Shiladitya Banerjee, School of Physics, Georgia Institute of Technology
Boundary Manifestations of Higher Berry Phases
Friday, November 6, 2026
10:00 a.m.
Howey W401
Virtual: Wenhao Qin Thesis Proposal
Committee Members:
Dr. JC Gumbart, School of Physics, Georgia Institute of Technology
Dr. Zeb Rocklin, School of Physics, Georgia Institute of Technology
Dr. Minsu Kim, Department of Physics, Emory University
]]>
Grace Barkhuff
Ph.D. Student in Human-Centered Computing
School of Interactive Computing
Georgia Institute of Technology
Date: November 13, 2026
Time: 3pm - 5pm ET
Location: TSRB 233
Committee
Dr. Ellen Zegura (Advisor) - School of Interactive Computing and School of Computer Science, Georgia Institute of Technology
Dr. Rodrigo Borela - School of Computing Instruction, Georgia Institute of Technology
Dr. Betsy DiSalvo - School of Interactive Computing, Georgia Institute of Technology
Dr. Carl DiSalvo - School of Interactive Computing, Georgia Institute of Technology
Dr. Casey Fiesler - Department of Information Science, University of Colorado Boulder
Abstract
The technology industry develops pervasive, ubiquitous tools that impact both society broadly and our personal, daily lives. From social media's influence on elections to generative AI for emotional support, computing technologies and those who develop them have a profound impact on how we live. As a result, calls abound for new computing technologies to be developed "ethically" and "responsibly." For the past fifty years, academia's efforts in this realm have focused on computing ethics education, a subject within computing that teaches students to consider the societal impacts of the tools they design. Because this education is critical, we must pause to evaluate the state of our ethics education and its effectiveness and imagine how we might improve it. In this thesis proposal, I provide an overview of the current state of ethics education in the United States, including an analysis of educational materials that meet ethics learning outcomes, educators' perceptions of ethics instruction, prevalence of required ethics instruction in computing programs, and students' approach to ethics when designing new technologies. I reflect on what it might mean for instructors to model ethical engagement with technology in their own classrooms, then propose work related to using ethics education to help students develop an understanding of how their values connect to their careers in computing.
Meeting Details
https://gatech.zoom.us/j/4265896666?pwd=bW10czF5UGVJcTNzWEFQa0NYem1TZz09
]]>
Advisor: Dr. Dimitri Mavris
Milestone: PhD Thesis Final Examination (Defense)
Degree Program: Aerospace Engineering
Title: A Scalable and Efficient Framework for Graph-Based Analysis of the Digital Thread
Abstract: The increasingly connected and rapidly evolving aerospace operational environment, coupled with tight budgets and aggressive timelines, poses significant challenges to the current document-centric analytical paradigm. Addressing these challenges requires Digital Transformation and the application of Digital Engineering principles, with the Digital Thread serving as a key enabler. However, the Digital Thread’s scale and complexity can limit the efficiency of advanced analytics. Digital Thread graphs can support these analyses through graph representation learning, particularly Graph Neural Networks (GNNs). Processing entire graphs introduces scalability challenges, while partitioning them into smaller isolated subgraphs restricts information flow. Furthermore, learning representations for prediction does not, by itself, provide an effective strategy for design-space exploration. This thesis addresses these challenges through hybrid graph representations and gradient-based exploration of regression embeddings. Hybrid graphs preserve detailed local information while incorporating aggregated global context from external partitions. Embedding space exploration searches learned representations and decodes them into admissible input states. The AC Optimal Power Flow (AC-OPF) problem in electricity transmission serves as the primary proxy for Digital Thread analytics. The methods are also applied to flight delay prediction and schedule optimization to assess whether the observed trends hold in a different application domain. Across both applications, hybrid graphs recover predictive accuracy lost through isolation while reducing peak memory requirements relative to processing entire graphs. This follows the expected balance between information retention and computational efficiency. Training time benefits are more conditional: additional training batches limit savings in the electricity transmission experiments, whereas hybrid models reduce training time in the flight delay demonstration without subsampling. Embedding space exploration also reduces search time and memory compared with optimizing the original system inputs directly. The improvement achieved in the learned representation must be translated/decoded back into feasible changes to the engineering system. Decoding preserves a substantial portion of the improvement obtained through direct input optimization in the electricity transmission experiments, with a greater loss of improvement in the flight schedule application. These findings show the potential and limitations of hybrid graphs and learned representations for computationally efficient graph-based analysis and optimization of engineered systems, providing a foundation for their application to Digital Thread graphs.
Date and time: 2026-11-16, 8:30 AM
Location: Collaborative Visual Environment (CoVE); Weber Space Science and Technology Building (SSTIII)
Committee:
Dr. Dimitri Mavris (advisor), School of Aerospace Engineering
Dr. Kyriakos G. Vamvoudakis, School of Aerospace Engineering
Dr. Kai A. James, School of Aerospace Engineering
Dr. Woong Je Sung, School of Aerospace Engineering
Dr. Arun Ramamurthy, Siemens Technology (Princeton
Committee:
Dr. Zonouz, Advisor
Dr. Mertoguno, Chair
Dr. Chou
]]>Committee:
Dr. Krishna, Advisor
Dr. Mahajan, Chair
Dr. Hao
Dr. Huang
Dr. Wang
]]>Committee:
Dr. Yu, Advisor
Dr. Lin, Chair
Dr. Mukhopadhyay
]]>
Doctor of Philosophy in Applied Physiology
In the
School of Biological Sciences
Joelle Dick
Will defend her dissertation
MANAGING REDUNDANCY FOR STABLE TASK PERFORMANCE UNDER CONSTRAINTS
Friday, November 6th, 2026
12:00 PM
Applied Physiology Building (555 14th St.), Room 1253
Thesis Advisor:
Young-Hui Chang, Ph.D.
School of Biological Sciences
Georgia Institute of Technology
Committee Members:
Lewis Wheaton, Ph.D.
School of Biological Sciences
Georgia Institute of Technology
Boris Prilutsky, Ph.D.
School of Biological Sciences
Georgia Institute of Technology
Greg Sawicki, Ph.D.
Schools of Mechanical Engineering and Biological Sciences
Georgia Institute of Technology
Colin Grove, Ph.D.
Department of Rehabilitation Medicine
Emory University School of Medicine
ABSTRACT:
Human movement requires the nervous system to use motor and sensory redundancy to maintain stability and achieve task-specific goals. This redundancy provides flexibility to adapt movement strategies in response to changing internal, mechanical, and environmental constraints. Understanding how the nervous system uses this flexibility to maintain task performance is essential for understanding motor coordination and adaptation.
This dissertation investigates how the nervous system organizes movement variability and joint coordination in response to three distinct constraints: neuromuscular fatigue, increasing running speed, and altered gravity. Using uncontrolled manifold analysis and system identification techniques, three studies examine how changing task demands influence joint-level coordination and the stabilization of task-relevant variables.
The first study investigated running speed in non-amputees and unilateral transtibial amputee sprinters, showing that runners without a prosthesis shift from interjoint toward individual joint coordination at higher speeds, whereas runners with a prosthesis relied more on interjoint covariation. The second study examined lower-limb coordination during repetitive squatting revealing that kinematic coordination remained relatively consistent as fatigue progressed. The third study examined postural control following simulated hypogravity adaptation, revealing a significant reduction in whole-body center of mass sway without significant changes in joint moment variability or relative ankle-to-hip contributions.
Together, these findings suggest that the nervous system responds to changing physiological, mechanical, and environmental constraints by adjusting movement coordination and control. This dissertation highlights the importance of interpreting movement variability in relation to task-specific goals and provides insight into how the human motor system uses redundancy and flexible coordination strategies to maintain performance across diverse conditions.
]]>
(Advisor: Prof. Vladimir V. Tsukruk)
will defend a doctoral thesis entitled,
Ionic and Interfacial Assembly of Inorganic–Organic Hybrid Nanomaterials with Chiral and Magnetic Functions
on
Tuesday, November 10, 2026,
at
11:00 a.m.–1:00 p.m.
in
Molecular Science and Engineering Building (MoSE) Room 1226
Abstract
Inorganic–organic hybrid nanomaterials with chiral and magnetic functions are promising for chiral photonics and enantioselective sensing, as well as for magnetically recoverable adsorbents and catalysts, because they combine the robustness of an inorganic component with the chemical tailorability of an organic one. However, these functions are usually installed through covalent synthesis or in organic solvents, and the resulting nanoscale building blocks are often fragile or metastable. How aqueous, non-covalent processing relates to the nanoscale structure and functional properties of these hybrids remains largely unexplored.
This work examines ionic pairing and interfacial coordination in water as processing routes to chiral and magnetic hybrid nanomaterials, and relates the resulting nanoscale structures to the chiroptical and magnetic functions of the hybrids. First, sulfonated polyhedral oligomeric silsesquioxane (POSS) nanocages were paired ionically with the amino acid L-tryptophan. The amino acid keeps its chirality on the rigid inorganic cage, while the cage makes the hybrid more robust and directs its assembly into ordered nanostructures. Second, a one-pot aqueous synthesis was developed in which solution basicity switches the shape of the layered zeolitic imidazolate framework ZIF-L, and magnetic nanoparticles attached to the ZIF-L surfaces add magnetic response and protect the metastable layered phase. Third, preliminary results combine the two approaches: magnetic nanoparticles leave the chirality of an ionic POSS–amino acid hybrid unchanged in solution and add in-plane order to its films, and POSS–nanoparticle hybrids can be organized by a magnetic field.
Together, these results identify the mobile ionic counterion and the nanoparticle surface as general design handles for functional hybrid materials made in water. This work provides a basis for the design of robust, water-processable chiral and magnetic nanomaterials, from chiral optical films to recoverable materials for water treatment.
Committee Members:
Prof. Vladimir V. Tsukruk (Advisor) – School of Materials Science and Engineering
Prof. Blair Brettmann – School of Chemical and Biomolecular Engineering
Prof. Scott Danielsen – School of Materials Science and Engineering
Prof. Shucong Li – School of Materials Science and Engineering
Prof. Todd Sulchek – George W. Woodruff School of Mechanical Engineering
Dr. Dhriti Nepal – Air Force Research Laboratory
]]>
(Advisor: Prof. Vladimir Tsukruk)
will defend a doctoral thesis entitled,
Engineered Nanoscale Interfaces for
Electrophysiological and Electrochemical Sensing
On
Thursday, November 12, 2026 at 9:00 a.m.
Krone Engineered biosystems building (EBB) 3029 Conference Room
and/or virtually via MS Teams
Abstract
The signal quality of a sensing device is largely determined at the interface between the device and its target, a region typically less than a few hundred nm thick. In wearable electrophysiology, this interface must maintain contact with skin that deforms and perspires, whereas in trace-level chemical sensing it must discriminate analytes with few reactive functional groups. In both cases, the thickness and surface chemistry of the interface are usually dictated by the choice of material. This reserach presents two nanoscale interface platforms in which both parameters are defined at the design stage.
The first platform is an ultrathin skin-conformal nano-electrode composed of Ti3C2Tx MXene (50 nm), poly-L-lysine, and parylene (≈300 nm), with the parylene thickness determined from a work-of-adhesion analysis of contact with skin texture. The hydrophilic MXene side and hydrophobic parylene side allow the film to be transferred onto skin by water-assisted capillary flow and to remain attached without gel or adhesive. Compared with commercial gel electrodes, the nano-electrode showed approximately 35% lower skin interfacial impedance and up to threefold higher signal-to-noise ratio in ECG and EMG. ECG was recorded in a sauna and a swimming pool, where gel electrodes detached within minutes, and stable recording was maintained over 30 h of continuous wear.
The second platform is an extended-gate field-effect transistor (EGFET) for tertiary amines of the tryptamine family, with gramine as a structural surrogate for N,N‑dimethyltryptamine (DMT). Since these molecules have no N–H bond and no useful redox activity, selectivity was introduced through chemical activation. Sequential N‑oxidation and Polonovski rearrangement convert the tertiary amine into an amide, which is covalently bound to an amino-terminated self-assembled monolayer on the gold gate by transamidation, and the resulting surface-potential change is read by a commercial MOSFET. The sensor exhibited a limit of detection of 7 nM over four orders of magnitude. In artificial saliva, gramine was still activated in the presence of thiocyanate and urea, and the extraction step of the Polonovski reaction separated the amide from the matrix before surface coupling.
In both platforms, a function normally supplied by an added layer was instead built into the interface itself by controlling its thickness and surface chemistry: water tolerance at the electrode and molecular selectivity at the gate.
Committee
BioE PhD Proposal Presentation
11:00 am, Friday, November 13th, 2026
Location: Whittaker 1103
Zoom Link: https://gatech.zoom.us/j/93160060109?pwd=bE94BQ44D5XKOHXVYgQ7UazBPpF4u0.1
Advisor: Dr. Wilbur Lam (Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University)
Committee Members:
Dr. David Myers (Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University)
Dr. Tara Deans (Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University)
Dr. Todd Sulchek (George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology)
Dr. Ovijit Chaudhuri (Department of Mechanical Engineering, Stanford University)
Dynamics of Blood Clot Viscoelasticity Mediates Clot Maturation via Platelet Mechanotransduction
Clot formation is an essential step in hemostasis and needs to be precisely regulated. Dysfunctional clot growth can manifest in different cardiovascular conditions. Excessive clotting can lead to thrombosis, while insufficient clotting can cause hemorrhagic disorders. While clot initiation has been extensively studied and well-characterized, clot cessation represents a critical gap in our understanding of the hemostatic process. Clots represent a highly dynamic environment, rapidly transitioning from a liquid-like, viscous material to a solid-like elastic material. Platelets have been known to sense their mechanical environments, specifically substrate stiffness. I extend this reasoning to substrate viscoelasticity and hypothesize that clot viscoelasticity is a mechanical regulator of clot growth arrest. To test this, I will use tunable interpenetrating networks (IPNs) to investigate how substrate viscoelasticity regulates platelet activation at the molecular level, and how physiological shear stresses interact with clot mechanics to govern clot cessation. Aim 1 will establish physiological viscoelastic relevance of the IPNs by comparing their viscoelastic properties of the IPNs to whole blood clots and contrived spatial clot models. Aim 2 will be to investigate platelet response to viscoelastic microenvironments in both 2D and 3D environments and elucidate the molecular mechanisms that govern platelet activation through integrin-driven signaling pathways. Aim 3 will extend these findings to physiological flow conditions using microfluidic platforms, determining the synergistic effect of hemodynamic shear and clot viscoelasticity on platelet mechanotransduction. These studies will uncover viscoelastic mechanosensing as a previously unrecognized regulator of clot cessation, generating a new mechanobiological perspective for understanding hemostasis and thrombosis with direct implications for the development of precision therapies for both thrombotic and hemorrhagic disorders.
]]>
Advisors: Josh Kacher and Aaron Stebner
will defend a doctoral thesis entitled,
Influence of Heterogeneous Particulate Reinforcements on Deformation and Damage in Aluminum Composites Formed by Reactive Additive Manufacturing
On
Thursday, November 12, 2026
12 pm – 2 pm
-------------
or virtually via Teams:
Join: https://teams.microsoft.com/meet/248266455555984?p=93WUJ1utqZm4Pd8eD9
Meeting ID: 248 266 455 555 984
Passcode: zZ9Kw7aL
Committee:
Prof. Josh Kacher - School of Materials Science and Engineering (advisor)
Prof. Aaron Stebner - School of Materials Science and Engineering (co-advisor)
Prof. Naresh Thadhani - School of Materials Science and Engineering
Prof. Richard Neu - Woodruff School of Mechanical Engineering
Prof. Ashley Spear - Department of Mechanical Engineering, University of Utah
Abstract
Laser powder bed fusion (PBF-LB) enables the fabrication of complex parts, but aluminum alloys manufactured through this process remain susceptible to solidification cracking and anisotropic mechanical properties associated with columnar grain growth. Reactive additive manufacturing (RAM) is a method of promoting grain refinement and introducing reinforcement particles through in situ chemical reactions between reactive powder constituents during processing. However, incomplete reactions contribute to heterogeneous microstructures containing variations in particle size, shape, composition, and spatial distribution, which can influence deformation and fracture behavior. Understanding these effects is important for identifying potential damage initiation mechanisms and their effect on tensile properties. In this work, I investigated the relationships between heterogeneous reinforcement particles, deformation and fracture, and mechanical properties in Al-Ti-B₄C composites formed through RAM.
First, in situ scanning electron microscopy (SEM) tensile testing was coupled with digital image correlation (DIC) to analyze local strain distributions as they relate to the reinforcement particles over different volume fractions. Machine learning approaches were then employed to quantify the relative importance of particle geometry and composition in predicting particle crack initiation and propagation under uniaxial tension. Finally, the laser-material interaction time during PBF-LB was progressively increased to study the influence of increasing thermal exposure on the progression of the in situ chemical reactions, porosity, and subsequent tensile properties. These studies provide a foundation for the microstructural considerations necessary in developing PBF-LB processing strategies that balance the benefits of in situ reinforcement particle formation against the introduction of features which may compromise tensile performance.
]]>Himadri S. Pandey
Machine Learning PhD Candidate
H. Milton Stewart School of Industrial and Systems Engineering
Georgia Institute of Technology
Committee
Abstract
Machine learning has shown great promise for improving healthcare through more accurate diagnosis, risk prediction, and personalized care. However, accurate predictions alone do not guarantee better decisions. In practice, healthcare providers must make decisions with limited time, staffing, and resources. Yet these constraints are often overlooked during model development. This dissertation investigates how optimization can be integrated into machine learning to develop decision-making tools that explicitly account for these operational constraints. It addresses three interconnected questions: When is individualized information worth collecting? What information should we prioritize when resources are limited? And who should receive scarce interventions, and when?
The first part of this dissertation examines the value of information in clinical decision-making, developing a framework to quantify how individualized baseline testing changes diagnostic decisions in sports-related concussion assessment. The second part develops an operationally constrained machine learning framework that jointly selects diagnostic assessments and trains interpretable prediction models while accounting for uncertain administration times, clinical knowledge, and operational requirements. The third part introduces a mixed-integer optimization framework for ranked feature selection that optimizes feature rankings simultaneously across multiple resource budgets, accounting for feature acquisition costs and the predictive value of each successive feature set. The final part addresses the dynamic allocation of scarce healthcare resources among pediatric patients whose clinical conditions evolve over time. It proposes a multi-action restless multi-armed bandit framework and a decision-focused learning approach that combines neural networks and long-run resource planning to coordinate clinical interventions under resource constraints.
Together, these contributions advance a shift in healthcare AI from “What can we predict?” to “What actions should we take and for whom, given the resources available?”
]]>School of Psychological and Brain Sciences
Ph.D. Dissertation Defense Meeting
Time: Monday November 9th, 2026 9am Eastern Time
Location: Room 150 or https://gatech.zoom.us/j/94024313801
Dissertation Committee Chair/Advisor:
Dr. Scott Moffat (Georgia Tech)
Dissertation Committee Members:
Dr. Robert Ariel – (Virginia Wesleyan University, Department of Psychology)
Dr. Thackery Brown – (Georgia Tech)
Dr. Paul Verhaeghen – (Georgia Tech)
Dr. Steven Weisberg – (University of Texas at Arlington, Department of Psychology)
Title: Individual Differences in Navigation Performance and Metacognitive Monitoring:
The Role of Confidence in Route Learning
Abstract:
Learning an environment requires both knowledge of spatial relationships and an ability to judge when that knowledge is reliable. Confidence may also influence the strategies navigators use both when learning an environment and retrieving spatial information. Seventy-one mainly younger adults learned landmarks on two routes in immersive virtual reality. They then made 56 directional pointing judgments, each accompanied by a chosen continuous angular area intended to contain the correct direction with 90% confidence. Based on pointing performance, participants were classified as integrators (accurate within and between routes), non-integrators (accurate within routes but inaccurate between routes), or imprecise navigators (inaccurate on both). They also completed baseline measures and a three-arm Y-maze to assess allocentric versus egocentric strategy preference. Pointing errors and confidence intervals were wider for between-route judgments. Absolute calibration gaps were larger between routes, with non-integrators resembling integrators within routes but imprecise navigators between routes. Relative monitoring was also substantially weaker between routes: a 10° increase in confidence half-width was associated with a 12.9% increase in pointing error within routes, compared with only 1.4% between routes. Integrators reported a better general sense of direction than imprecise navigators. Interestingly, non-integrators performed relatively well on mental rotation and reported comparatively strong spatial ability, yet rated their general sense of direction more poorly than integrators. These findings suggest that confidence reflected the greater demands of between-route judgments but provided less information about the accuracy of individual responses. Non-integrators' relatively strong spatial abilities and within-route performance suggest that their difficulties with route integration may reflect how spatial knowledge is constructed or applied, rather than a general limitation in spatial ability. Testing whether confidence influences strategy selection, rather than simply reflecting performance, could help explain why some navigators with otherwise adequate spatial abilities seem to struggle to integrate routes into a survey representation of the environment.
]]>The Role of Confidence in Route Learning
]]>
Advisor: Dr. Lakshmi Sankar
Milestone: PhD Thesis Final Examination (Defense)
Degree Program: Aerospace Engineering
Title: Roughness Tolerant Airfoil Design for Rotor and Propeller Applications
Abstract: An inverse design methodology for airfoil was used to design roughness tolerant airfoil for subsonic applications in propeller and helicopter rotor. Methodologies for designing airfoils were first validated with known airfoils such as the NACA airfoils, and were subsequently used for designing a family of airfoils with various thickness to chord ratios from 10% to 22% for UAV propeller and helicopter main rotor improvement applications. While designing the roughness tolerant airfoils, pressure distribution shape on the airfoils was chosen so that the adverse pressure gradient was delayed as far back on the airfoil on the upper surface as possible to delay boundary layer transition and reduce boundary layer thickness. Drag polars of smooth and rough airfoils between designed and baseline airfoils were compared and shown better drag buckets for a range of useable lift coefficients prior to propeller and rotor applications. Icing simulation was also completed to demonstrate lower ice accumulation (a natural roughness phenomenon) for the newly designed airfoils. In-house blade element-momentum and CFD tools were used to analyze propeller and helicopter rotor performance in 3D applications. The current inverse design methodology was able to successfully demonstrate a few possible applications where the airfoils are more roughness tolerant than nominal airfoils. Applications that used designed airfoils showed slightly better performance at higher thrust and lift conditions compared to nominal propeller or helicopter rotor applications. An undesirable higher nose down pitching moment was introduced with the new airfoil design due to aft shifting of loads to delay transition and reduce drag. It was observed that the new airfoils are tolerant to lower surface roughness only. Upper surface roughness tolerance still requires more investigation.
Date and time: 2026-11-30, 10am-12pm
Location: MK317
Committee:
Dr. Lakshmi Sankar (advisor), School of Aerospace Engineering
Dr. Brian German, School of Aerospace Engineering
Dr. Suhas Jain, School of Mechanical Engineering
Dr. Graeme Kennedy, School of Aerospace Engineering
Dr. Juergen Rauleder, School of Aerospace Engineering
Advisor: Dr. Keegan Moore
Milestone: MS Thesis Proposal
Degree Program: Aerospace Engineering
Title: A Physics-Enforced Dual-Phase Framework for Dynamics Identification in Forced Nonlinear Systems
Abstract: System identification is critical to many engineering applications, especially those involving vibrating structures. The Energy-based Dual-phase Dynamics Identification (EDDI) method is a data-driven technique that discovers equations of motion from measured responses in compliance with physical balance laws. This work expands the original EDDI method to the identification of nonlinear single-degree-of-freedom (SDOF) oscillators under external forcing. The major innovation presented here is the reformulation of EDDI to incorporate external forces into the identification process. The method is deployed in two phases to systematically extract mathematical models for the damping and stiffness of the structure. At its core, EDDI leverages conservation of energy and Newton’s second law to express the measured data in a form that allows identification of the underlying dynamics. Phase I estimates the energy dissipated by damping using conservation of energy, and then uses that estimate to identify a mathematical model for the damping. Phase II applies Newton’s second law to isolate the conservative force directly, which is then used to identify a mathematical model for the stiffness of the structure. The method is first demonstrated using a numerically simulated Duffing oscillator with known parameters to test the effectiveness of the proposed method. It is then applied to experimentally measured responses of a nonlinear oscillator with unknown stiffness and damping parameters, excited by a nonlinear external force through a magnetic stinger and a modal shaker. EDDI requires only the measured response, the external force, and the mass of the oscillator, and no prior information about the system dynamics is necessary
Date and time: 2026-11-11, 9am-11am
Location: Weber 200 Conferce Room
Committee:
Dr. Keegan Moore (advisor), School of Aerospace Engineering
Dr. Graeme Kennedy, Associate Professor ,School of Aerospace Engineering
Dr. Alper Erturk, Professor, School of Mechanical Engineering
,
School of Psychological and Brain Sciences – Ph.D. Dissertation Defense Meeting
Date: Friday, November 13th, 2026
Time: 9:00 - 11:00 AM
Location: JS Coon 148 or online (https://teams.microsoft.com/meet/2176377007980978?p=agsRjcSoRb5kP54DnK)
Dissertation Committee Chair/Advisor:
Bruce Walker, Ph.D. (Georgia Tech)
Dissertation Committee Members:
Richard Catrambone, Ph.D. (Georgia Tech)
Mark Himmelstein, Ph.D. (Georgia Tech)
Eric Schumacher, Ph.D. (Georgia Tech)
Nina Dalkner, Ph.D. (Medical University Graz)
Title: Development and Validation of the Model of Acceptance of Futuristic Technologies (MAFT)
Abstract: Established models of technology acceptance (i.e., TAM, UTAUT) provide parsimonious explanations of adoption when users can evaluate technology through experience or familiar comparisons. They are less explicit about how adoption intentions form when a technology is not yet available, its operation and consequences are difficult to understand, and reliable evidence concerning performance, failure, or misuse is limited. This dissertation developed and tested the Model of Acceptance of Futuristic Technologies (MAFT) to explain early adoption intentions under these conditions of experiential and informational uncertainty. Across three studies, flying cars were used as an exemplar futuristic technology. The research program combined assessments of prior beliefs and discrete emotional responses, structural equation modeling, and pre-post simulated exposure delivered through desktop-based 360 video and immersive virtual reality. The findings indicate that prior beliefs and affective responses provided an initial interpretive framework through which perceived value, trust, and risk were evaluated. These judgments formed an affectively organized pathway associated with adoption intention, with perceived risk playing a less dominant role than originally anticipated. Although psychological distance contributed to the broader framework, the findings placed greater explanatory emphasis on affective and evaluative processes. MAFT extends technology acceptance research by explaining how people evaluate emerging technologies before meaningful experience and reliable performance evidence are available. The findings suggest that responsible communication, design, and governance should support appropriately calibrated beliefs, affective expectations, and trust by presenting realistic benefits, limitations, uncertainties, safeguards, and error modes. Such calibration may reduce both overtrust and unwarranted rejection of technological novelty.
]]>Doctor of Philosophy in Physics
School of Physics Thesis Dissertation Defense
Megan Arogeti
Dr. Laura Cadonati, School of Physics, Georgia Institute of Technology (Advisor)
Gravitational Wave Signal Reconstruction and Binary Neutron Star Postmerger Detection Using Stacking Methods
Date: Thursday, November 5, 2026
Time: 12:00 p.m.
Location: CRA Seminar Room (College of Computing) CCB 103
Virtual: https://teams.microsoft.com/meet/240561325650301?p=nvL8ezZRRym7CDoxSV
Thesis Committee:
Dr. David Ballantyne, School of Physics, Georgia Institute of Technology
Dr. Gongjie Li, School of Physics, Georgia Institute of Technology
Dr. John Wise, School of Physics, Georgia Institute of Technology
Dr. Edoardo Milotti, Physics Department, University of Trieste
Abstract:
September 14, 2015 marks the date of the first direct observation of a gravitational wave (GW). In the ten years that followed, the LIGO-Virgo-KAGRA (LVK) collaboration has observed hundreds of signals from merging compact binaries. Beyond measuring the extremely small distortion of spacetime caused by a passing gravitational wave, detecting and interpreting these signals requires advanced techniques. In this talk, I discuss two aspects of GW data analysis: validating the reliability of modeled waveforms and pushing our detection sensitivity beyond the limits permitted by individual events. I begin with a review of gravitational-wave astronomy and LVK observations. I then describe a method where we compare a source-agnostic reconstruction of a gravitational wave signal to a best-fitting waveform from a template-based analysis used to infer the source's properties. This test allows us to validate the efficacy of our waveform models and establish whether they are able to accurately describe the data. I report the results of my work applying this analysis to events from the fourth LVK observing run. In the second part of this talk, I first review the gravitational wave signals produced by a deformed binary neutron star (BNS) postmerger remnant. BNS postmerger signals carry information about the dense matter within neutron star cores, and could, therefore, be a direct probe into the properties of matter under some of the most extreme conditions. However, these signals are expected to be emitted at frequencies between 1-4 kHz, where detector sensitivity is reduced, making them extremely difficult to observe. For the next LVK observing run, the probability of a confident individual postmerger detection is well below one per year. With detector improvements, and especially third-generation detectors, we expect an increase in the number of BNS inspiral detections, some of which may contain postmerger emission that is too weak to be detected on its own. I therefore introduce a method for combining signals from multiple events in their time-frequency space to enhance postmerger detection probabilities. I then describe a framework to determine the presence of a BNS postmerger signal and to measure its characteristic frequency. Finally, I present the results from a proof-of-concept simulation study using simulated signals from three models of neutron star matter. I demonstrate that combining events has the power to increase the fraction of postmerger signals detected compared to analyzing events individually, showing promise for future observations.
]]>
Date: Tuesday, November 17, 2026
Time: 12:00 PM to 2:00 PM Eastern Time (US)
Join in Person: CODA 114 (first floor conference room; just walk in, no special access needed)
Join Virtually: Zoom
ShengYun Anthony Peng
CS Ph.D. Candidate
School of Computational Science and Engineering
College of Computing
Georgia Institute of Technology
Committee:
Dr. Duen Horng (Polo) Chau — Advisor and Committee Chair, School of Computational Science and Engineering, Georgia Institute of Technology
Dr. Wei Xu — School of Computer Science, Georgia Institute of Technology
Dr. Alan Ritter — School of Interactive Computing, Georgia Institute of Technology
Dr. Jianfeng Chi — Member of Technical Staff, OpenAI
Dr. Pin-Yu Chen — Principal Research Scientist, IBM Research
Abstract:
Foundation models achieve remarkable benchmark performance, but growing capabilities do not guarantee safety, reliability, or efficiency in real-world deployment. My dissertation empowers foundation models to operate safely and reliably at scale through three complementary thrusts:
(1) Aligning for Safety. We discover a universal safety basin phenomenon across popular open-source LLMs: a local parameter region preserving safety. This discovery, featured in the International AI Safety Report, reveals why finetuning erodes alignment and inspires STAR-DSS, our novel dynamic safety shaping method to mitigate this erosion. For safer reasoning, we introduce RECAP, a principled reinforcement learning (RL) method for post-training that teaches models to recover from flawed thinking.
(2) Designing for Reliability. Safe behavior must be paired with reliable predictions. In Robust Principles, we synthesize a suite of three generalizable robust architectural design principles for CNNs, achieving state-of-the-art (SOTA) adversarial robustness on RobustBench. This work earned the Best Poster (Paper) Award at BMVC and guided DARPA’s research on adversarially robust architectures. These insights also enable accurate, efficient business table recognition with lightweight encoders.
(3) Scaling Model Capabilities. Complementing architectural design, we unlock accurate table recognition with simpler encoders through self-supervised pretraining. Our UniTable training framework unifies both the training paradigm and training objective of table recognition, achieving SOTA performance. UniTable (530+ GitHub stars) powers table extraction in Open Parse (3.2K GitHub stars) and is used by startups for retrieval-augmented generation (RAG). This research has been selected for oral presentations and presented in invited talks. To scale visual understanding efficiently, we establish a principled framework revealing optimal inference compute allocations for video VLMs.
My dissertation research significantly impacts academia, industry, and society, advancing foundation models through collaborations with NVIDIA, Meta, IBM, Intel, and ADP. It has contributed to the AI foundations of industry research grants totaling over $1.7 million.
]]>BioE PhD Defense Presentation
Date: Wednesday, November 11th, 2026
Time: 2:00 PM
Location: EBB CHOA Seminar Room
Virtual Meeting Link: https://gatech.zoom.us/j/91747576515
Committee Members:
Leslie Chan, PhD (Advisor); M.G. Finn, PhD; Andrés García, PhD; Rheinallt Jones, PhD; Gabe Kwong, PhD
Engineering ingestible probes for breath-based detection of gastrointestinal diseases
Gastrointestinal (GI) diseases are a significant healthcare burden and affect 70 million people in the United States. Endoscopic imaging plays a crucial role in the diagnosis and clinical management of many GI diseases, including GI cancers, gastroesophageal reflux disease, and inflammatory bowel disease (IBD). However, its invasive nature deters individuals from recommended testing and hinders its routine use as a monitoring tool. Non-invasive tests for detection of blood and fecal biomarkers are used in the clinic, but either have poor sensitivity for early detection or poor disease specificity. For these reasons, there remains a critical unmet need for non-invasive tests for GI disease.
Breath tests are a highly attractive alternative to endoscopy because breath can be sampled non-invasively and analyzed for disease-indicating volatile organic compounds (VOCs) (i.e., breath biomarkers). However, few breath biomarkers have been successfully identified due to technical and statical limitations of conventional -omic approaches. In this work, we have developed an approach to induce exhalation of breath biomarkers to diagnose and monitor GI diseases. Our strategy uses new ingestible molecular probes, which release volatile reporters upon activation by disease activity in the GI tract. Once released, reporters are rapidly cleared from the body in breath and measured via mass spectrometry. Thus, probe ingestion induces exhalation of pre-specified volatile reporters of known mass to indicate disease activity (i.e., synthetic breath biomarkers), obviating the need for biomarker discovery.
Our approach leverages dysregulated glycosidase activities, which play a central role in the pathogenesis of diverse GI diseases. The GI tract houses hundreds of distinct glycosidases from the host and microbiome that regulate gut barrier function, mucosal immunity, host-microbe interactions, and the metabolism of drugs and dietary sugars. In Aim 1, I establish design principles for the ingestible probes and develop a probe to monitor the activity of a microbial glycosidase that causes drug-induced GI toxicity. In Aim 2, I use the established platform to develop a probe for a human intestinal glycosidase which, when deficient, leads to a sugar malabsorption disorder. I further demonstrate how induced breath signals can be detected via novel carbon nanotube detection arrays, a low-cost alternative to mass spectrometry. Lastly, in Aim 3, I demonstrate the multiplexing capabilities of our platform and assemble a volatile-barcoded probe array to induce breath biomarker signatures of mucosal injury and healing in the context of IBD.
]]>
Time: 11:00 am
Room: Skiles 168
Title: On the bow lemma and higher dimensional analogs.
Committee: Igor Belegradek, Elena Braverman (University of Calgary), Mohammad Ghomi (advisor), Galyna Livshyts, Andrzej Swiech, Erik Verriest (ECE).
Abstract:
We discuss the bow lemma or Schur comparison theorem, and in particular its higher dimensional analogs. We first investigate isoperimetry under upper and lower curvature bounds, including the Cartan-Hadamard conjecture. We then discuss Fenchel's theorem for CAT(0) spaces and its higher dimensional versions concerning tight (hyper)surfaces in Cartan-Hadamard manifolds. We also develop a rigidity mechanism for flat hypersurfaces in Alexandrov ambient spaces, which is applied to completely characterize surfaces which optimize total Gauss curvature in 3-folds with upper or lower curvature bounds.
Teams link: https://teams.microsoft.com/meet/2172100941812051?p=0YimD0FGzxQfer6TrD
]]>
School of Psychological and Brain Sciences – Ph.D. Dissertation Defense Meeting
Date: Monday, November 16th, 2026
Time: 1:00 - 3:00PM
Location: Price Gilbert room 4222 or online (https://teams.microsoft.com/meet/252794986155875?p=9IvZNJbLpRMdqXnFqk)
Dissertation Committee Chair/Advisor:
Bruce Walker, Ph.D. (Georgia Tech)
Dissertation Committee Members:
Richard Catrambone, Ph.D. (Georgia Tech)
Woon Ju Park, Ph.D. (Georgia Tech)
Ben Satterfield, Ph.D. (Georgia Tech)
Barbara Chaparro, Ph.D. (Embry-Riddle Aeronautical University)
Title: From Acceptance to Mastery: An Evaluation of Technology Mastery and Validation of the General Technology Mastery Scale
Abstract: This dissertation empirically investigated the concept of technology mastery, operationally defined here as an individual’s knowledge, attitudes, and ability regarding the use of a technology. To begin, a series of interviews with a mix of community members and college undergraduates were conducted to better define the concept of “technology mastery,” determine what behaviors and traits are associated with the term, and finally discuss any barriers faced when attempting to gain skill with technology. The results of this study were thematically analyzed and produced roughly 40 themes, which were then used to create the General Technology Mastery Scale (GTMS) through a series of factor analysis techniques. The GTMS then underwent a series of validation testing through comparisons with hypothesized covariates based on the literature (such as problem-solving, technology self-efficacy, general self-efficacy, technology acceptance, system usability, locus of control, motivation, and creativity). Finally, the GTMS was deployed in a classroom setting to assess within-subjects growth in mastery, motivation, and problem-solving ability. This dissertation contributes a consolidation of the work done in technology mastery and fluency, in addition to providing validated psychometric measuring techniques to assess technology mastery. The information suggested by the results of this dissertation can be used to help establish a strong foundation for developing skilled technology use, thus ensuring that the benefits of technology are realized in every aspect of our lives.
]]>
Date: 10th Nov 2026
Time: 10:00 to 12:00 EST
Location: Virtual via Zoom — https://gatech.zoom.us/j/98959804477?pwd=Evx56C26tohvh3a6Q4HQzyA2861uAz.1 (Meeting ID: 989 5980 4477, Passcode: 070732)
Ayush Mohanty
Machine Learning PhD Student
H. Milton Stewart School of Industrial and Systems Engineering
Georgia Institute of Technology
Committee
1 Prof. Nagi Gebraeel (Advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
2 Prof. Pascal Van Hentenryck, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
3 Prof. Yao Xie, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology
4 Prof. Aaron Stebner, George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology
5 Prof. Jeff Shamma, The Department of Industrial and Enterprise Systems Engineering, University of Illinois Urbana-Champaign
Abstract
Modern industrial systems comprise multiple interconnected subsystems whose temporal dynamics are often interdependent, making it essential to understand how changes in one subsystem influence others. Learning these interdependencies is particularly challenging in decentralized environments where raw data cannot be shared, and individual subsystems operate proprietary models that cannot be modified. This dissertation develops decentralized learning frameworks for characterizing cross-client temporal interdependencies in both linear and nonlinear dynamical systems while preserving the autonomy of individual client models. For linear systems, we establish a federated Granger causality framework that enables the discovery of unknown cross-client interdependencies. We further use this federated Granger causality framework to investigate the propagation of estimation uncertainty and its implications for interdependency learning. For nonlinear systems, the dissertation introduces graph attention-based learning with a predefined candidate graph structure, establishing theoretical relationships between learned attention coefficients and Jacobian-based interdependencies. It further develops a complementary framework for learning nonlinear interdependencies without requiring a predefined graph skeleton, where the downstream task of root cause analysis is explored. The proposed methods are evaluated on synthetic and industrial datasets, demonstrating their utility for decentralized prediction and interdependency characterization.
]]>
Christopher Kraemer
PhD Candidate
School of Computer Science
College of Computing
Georgia Institute of Technology
Committee
Dr. Josiah Hester(Advisor) - School of Computer Science, Georgia Institute of Technology
Dr. Ellen Zegura - School of Computer Science, Georgia Institute of Technology
Dr. Umakishore Ramachandran, School of Computer Science, Georgia Institute of Technology
Dr. Alex Adams - School of Interactive Computing, Georgia Institute of Technology
Dr. Thomas Ball - Honorary Professor, Lancaster University & Affiliate Faculty, University of Washington
Abstract
Three pillars of democratization are equity, access, and agency. An equitable system lowers the barrier to entry for novice users and programmers. First, I show equity through Battery-free MakeCode, a compiler modification to the popular education tool Microsoft’s MakeCode for micro:bit. This compiler modification analyses, reorganizes, and inserts code into a novice’s program so that even someone who is just learning how to code can make a battery-free device with the BBC micro:bit. Second, access means expanding battery-less computing to hobbyists, makers, and researchers. In my paper User-directed Assembly Code Transformations Enabling Efficient Batteryless Arduino Applications, I developed a low-level binary rewriting tool to enable battery-free computing for Arduino users. This allows Arduino users to make an intermittent device with a wide array of off-the-shelf hardware. Third, education and feedback about how a battery-free device will run in the wild gives programmers agency in their development. The Intermittent Computing Energy Toolkit revisits Microsoft's MakeCode for micro:bit and overhauls its debugging and simulator environment with dynamic and static energy analysis. This analysis allows users to detect battery-free energy bugs and test their programs before deploying their device.
Chih-Yu Lo
Advisor: Dr. Xueda Wen, School of Physics, Georgia Institute of Technology
Boundary Manifestations of Higher Berry Phases
Monday, November 2, 2026
1:00 p.m.
Howey N110
Committee Members:
Dr. Brian Kennedy, School of Physics, Georgia Institute of Technology
Dr. Phillip First, School of Physics, Georgia Institute of Technology
Dr. Carlos Sá de Melo, School of Physics, Georgia Institute of Technology
]]>
Ph.D. Thesis Defense Announcement
Examining the Role Of Automation Technology in Transportation Interactions: How Vehicle Affordances Affect Equity and User Well-Being
By: Christian Douglas
Molecular Sciences and Engineering 1222
https://gatech.zoom.us/j/94927419491
2026-11-13 11:00 AM ET
Dr. Adjo Amekudzi-Kennedy, Dr. Sofía Pérez-Guzmán, Dr. David Ederer, Dr. Mark Tschaepe
Autonomous vehicles (AVs) and vehicle automation technology come in all shapes and forms. Vehicle producers are including advanced technology in new vehicles, and ride-hailing companies are deploying AVs in cities across the United States. As such, it is crucial to look across all levels of automation to evaluate the usability of vehicle automation technology and AVs. This research examines the ability of transportation features and technology to provide affordances (i.e., aspects of the environment providing functionality) to a diverse set of potential users. This work employs a variety of methodological approaches to answer the central research question of how vehicle automation technology provide affordances to human drivers. This mixed-methods, interdisciplinary research addresses this fundamental and guiding research question by (1) advancing a disciplinary understanding of equitable transportation technology; (2) gathering the insight and experience of vehicle users, designers, and regulators; and (3) utilizing a hypothetical lane reduction scenario to explore the potential affordances, and disaffordances, resulting from interactions between vehicles at varying levels of automation. Findings can be used to advise vehicle producers on inclusive vehicle design practices, inform policy guiding nascent vehicle technology, and help future researchers understand the direction these new technologies are going and the threats they can pose to individual safety and well-being.
]]>Title: Modeling crime response to deterrence and positive incentives: existence of solutions, optimal policies, and fairness
Date: November 13th , 2026
Time: 11 AM - 1 PM, EST
Location: Guggenheim 246
Zoom link: https://gatech.zoom.us/j/7854428648?pwd=MjkvcHRXY2lqZVJXRS9PS1Y4dEVkZz09
Committee:
Dr. Martin Short (Advisor)
Dr. Sung Ha Kang
Dr. Wenjing Liao
Dr. Haomin Zhou
Dr. George Mohler
Abstract:
In this work we study a model in which rational agents decide whether or not to commit crime based on a utility calculation, influenced by a judge who sets a society-wide threshold corresponding to the likelihood of an individual being found guilty and a legislator who sets a society-wide punishment level. We first study how the overall crime rate is influenced by the judge's threshold and the legislator's punishment level, propose an objective function for the judge and legislator to minimize, and study the optimal threshold and punishment levels for this objective. We then consider the case in which the overall society is subdivided into multiple groups with varying characteristics, introducing a constraint on fairness in treatment between the groups. We study how an optimal threshold and punishment level might be chosen under this fairness constraint, what ramifications the constraints have on outcomes for individuals, and under what circumstances the constrained optimum agrees with the unconstrained optimum. Then we introduce a new lever for the justice system in the form of positive incentives for the agents: first as a subsidy and then later as a program motivated by the Trusted Traveler Program where agents can opt in or opt out from taking the positive incentive in exchange for some extra initial investigation. We then study the optimal threshold, punishment level, and positive incentive level that achieves the justice system's objective for both unlimited budget and a budget constraint
]]>
You are cordially invited to my PhD thesis defense on Monday, November 23rd.
Title: Adaptive and Robust Model Fusion for Multi-Modal Learning Systems
Date: Monday, November 23, 2026
Time: 1:00 PM - 3:00 PM Eastern Time (US)
Location: KACB 3402
Virtual (Zoom): https://gatech.zoom.us/j/94537626884?pwd=f2aDUdkbH2u8p6w9oLP0zMBdv6CJCN.1
Selim Furkan Tekin
Ph.D. Candidate in Computer Science
School of Computer Science
Georgia Institute of Technology
Committee members:
1. Dr. Ling Liu (Advisor) - School of Computer Science, Georgia Institute of Technology
2. Dr. Francisco Romero - School of Computer Science, Georgia Institute of Technology
3. Dr. Anand Padmanabha Iyer - School of Computer Science, Georgia Institute of Technology
4. Dr. Calton Pu - School of Computer Science, Georgia Institute of Technology
5. Dr. Margaret Loper - Information and Communications Laboratory, Georgia Tech Research Institute
6. Dr. Lakshmish Ramaswamy - School of Computing, University of Georgia
Abstract
The advancement of large-scale AI models in recent years has shown their remarkable performance across a broad spectrum of zero- and one-shot tasks across language and vision modalities. However, based on the benchmark studies and model release announcements revealed that even the newest released or largest models tend to carry some weaknesses despite the unique strengths, indicating no one model can be the persistent winner in generalization performance across all tasks. Complementing this, user-generated queries tend to have varying complexity and involve simple common knowledge tasks or complex reasoning or specific domain knowledge. For simple queries involving superficial understanding, smaller models can do a perfect job at a minimal cost. Hence, there is a challenge in adequately choosing a model or group of models based on the client-user constraints and the downstream learning tasks at hand. This dissertation is dedicated to tackling the above grand challenges by designing and developing innovative methodologies, algorithms, and optimizations in selecting, refining, and integrating multiple gen-AI models.
Our research began with a comprehensive empirical study, which led to three observations: First, in classical few-shot learning methods, ensemble gains come largely from two channels: structural (model architecture) and geometric (distance metric) diversity–motivating heterogeneous model selection. We later found that the same holds for modern few-shot learners such as LLMs and VLMs. Second, larger heterogeneous ensemble teams may not necessarily offer better generalization performance, because some members make redundant predictions that add inference cost without adding new information. Third, the language models and vision models require different multi-stage representation learning to effectively capture duplicated learning capabilities. Motivated by these observations, we have made four novel and fundamental contributions:
First, we have developed a novel diversity computation framework that can be tailored for both language and vision representation learning to capture the disagreement correlation among multiple gen-AI models (LLMs, LVMs, and VLMs).
Second, we have designed the first of its kind diversity-optimized ensemble pruning framework, which supports LLM and VLM ensemble pruning. The pruning utilizes genetic search optimization on introduced metrics: the focal-diversity metric and focal Central Kernel Alignment (CKA) to identify informative complementary representations for vision-language ensemble pruning.
Third, we study learning-to-combine architectures and develop two diversity-optimized ensemble fusion methods. The first is a supervised approach that uses diversity-driven pruning to select an ensemble team offline and then learns to fuse its outputs, achieving stronger generalization than state-of-the-art methods. The second selects the team and fuses outputs dynamically at inference time, using a two-stage reinforcement learning framework.
Finally, we address model routing and fusion through joint optimization, learning how to select, reason, and refine across multiple VLMs. We develop two routing systems trained with reinforcement learning. The first starts from prior beliefs about each VLM’s performance: a routing policy chooses which model to consult, and a fusion policy decides whether to keep consulting or answer. The second targets physical reasoning in real-world scenes: it matches the skills a query requires to each VLM’s perception strengths, routes to models in sequence, and refines a lightweight policy using critiques from a tool-augmented critic that extracts visual clues. In both lines of work, we compare our approaches against widely used AI chatbots and show that our approaches succeed in real-world scenarios where existing models often fail.
Collectively, these contributions offer new ways to build teams of diverse AI models that reason together, combine their outputs, and verify their answers. Such methods become increasingly important as the number and variety of large-scale models continue to grow.
]]>
Ph.D. Thesis Defense Announcement
Advancing the Practical Integration of Attitudes into Regional Travel Demand Models: Attitudinal Marker Variables, Systematic Model Specification, and Temporal Stability of Attitudes
By Ilsu Kim
Advisor:
Dr. Patricia Mokhtarian
School of Civil and Environmental Engineering
Georgia Institute of Technology
Committee Members:
Dr. Michael Hunter
School of Civil and Environmental Engineering
Georgia Institute of Technology
Dr. Sofia Perez-Guzman
School of Civil and Environmental Engineering
School of Industrial and Systems Engineering
Georgia Institute of Technology
Dr. Subhrajit Guhathakurta
School of City and Regional Planning
Georgia Institute of Technology
Dr. Ram Pendyala
School of Sustainable Engineering and the Built Environment
Arizona State University
Date and Time: Monday, November 9, 2026, at 3:00 PM EST
Location: Mason 2119 or online via Microsoft Teams
ABSTRACT
Facing evolving work and commute patterns, shifting lifestyles, and
emerging travel options such as micromobility and self-driving vehicles, travel
demand models (TDMs) need, and will continue to need, enhanced predictive
performance. Transportation-related attitudes help explain why individuals and
households with similar socioeconomic, demographic, and residential
characteristics make different travel choices. However, integrating attitudes into
regional TDMs requires practical ways to measure them, incorporate them into
travel behavior models, and forecast their shifts over time.
The goal of this dissertation is to support the practical integration of
traveler attitudes into TDMs. The marker variable (MV) approach reduces
measurement burdens by using a small set of selected attitudinal statements.
Attitudinal MVs can be used to impute attitudinal factor scores through functions
trained on an attitude-rich donor dataset or included directly in travel behavior
models. The dissertation addresses three complementary questions: (1) What are
the types and magnitudes of benefits obtained from incorporating MV-predicted
attitudinal factor scores or MVs into various travel behavior models? (2) How can
attitudinal MVs be best leveraged to improve the performance of individual travel
behavior models within regional TDMs? and (3) How stable are transportationrelated
attitudes over time, and what factors are associated with attitudinal shifts?
The findings show that including attitudes yields meaningful benefits,
particularly in predicting less-common alternatives, and can reveal additional
significant non-attitudinal explanatory variables. Segmenting the population and
incorporating attitudes into latent class membership further improve vehicle
ownership models, especially for single-adult households, highlighting the value of
allowing model structures and attitudinal roles to vary across segments.
Attitudinal stability differs across domains, with the pro-car-ownership attitude
showing greater persistence than the pro-transit and pro-active-mode attitudes,
while within-person shifts remain largely unexplained by observed characteristics
and their changes. Overall, this dissertation supports the practical integration of
attitudes into regional TDMs by demonstrating the value of abbreviated
measurement, offering a structured model specification process, and providing
insights into temporal stability. These findings inform how attitudes can be
collected and used in modeling practice, while identifying further research needed
to assign attitudes to individuals and forecast their evolution over time.
BioE PhD Defense Presentation
10-11am on Friday, November 6th, 2026
Location: IBB Suddath Meeting Room #1128
(Meeting ID: 935 505 593 46)
Committee:
Andrew J. Feola, PhD (Advisor)
(Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University; Department of Ophthalmology, Emory University; Atlanta VA Medical Center for Visual & Neurocognitive Rehabilitation)
C. Ross Ethier, PhD
(Department of Ophthalmology, Emory University, Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University)
Todd Sulchek, PhD
(Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University; George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology)
Levi Wood, PhD
(Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology & Emory University; George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology)
W. Daniel Stamer, PhD
(Department of Ophthalmology, Duke University)
Early-onset menopause impact on the biomechanics and transcriptomic profile of the conventional outflow pathway
Glaucoma, the leading cause of irreversible blindness worldwide, is characterized by peripheral vision loss due to retinal ganglion cell (RGC) loss. It’s projected to affect 112 million people by 2040, and 60-67% of the glaucomatous population are women. While elevated intraocular pressure (IOP) remains a major risk factor associated with developing glaucoma, recent work suggests that menopause is a sex-specific factor associated with glaucoma.
Studies have found that menopause had a linear association with the timing of glaucoma diagnosis, and that early-onset menopause increases the risk of developing glaucoma later in life. Menopause has also been shown to increase intraocular pressure (IOP) – the only modifiable risk factor for glaucoma. However, the role of menopause in glaucoma development is still not well understood. To address this gap, this work characterizes the potential impact of early-onset menopause on IOP, eye biomechanics, and transcriptomic changes within the conventional outflow pathway, which is largely responsible for regulating IOP.
In a preclinical approach we modeled early-onset menopause via ovariectomy (OVX). In Aim 1, we focused on determining the impact of early-onset menopause over time on IOP and aqueous humor outflow facility, which is how easily fluid can exit the eye. In Aim 2, we investigated the impact of early-onset menopause on the stiffness of the segmental flow of aqueous humor (low flow and high flow) through the trabecular meshwork (TM), the major tissue regulating aqueous humor outflow within the conventional outflow pathway. Finally, Aim 3 assessed the transcriptomic changes within the TM induced by early-onset menopause to elucidate potential transcriptional mechanisms driving these functional changes. This work fills an existing gap in the field regarding the effects of early-onset menopause on parameters associated with glaucoma. Thus, in addition to potentially elucidating novel components of glaucoma etiology, this work has the potential to identify therapeutic targets that can lead to new glaucoma treatments in the clinic.
]]>
(Advisors: Prof. Scott Danielsen, Prof. Jason Azoulay, and Prof. Natalie Stingelin)
Will propose a doctoral thesis entitled,
Post-Polymerization Modification of Aromatic Polymers and their Blends
On
Friday, November 6, 2026 from 12:30 pm – 2 pm ET
Molecular Science and Engineering Room 1224
Committee
Prof. Scott Danielsen, School of Materials Science and Engineering (Advisor)
Prof. Jason D. Azoulay, School of Chemistry and Biochemistry, and Materials Science and Engineering (Advisor)
Prof. Natalie Stingelin, School of Materials Science and Engineering (Advisor)
Prof. Will Gutekunst, School of Chemistry and Biochemistry
Prof. Rampi Ramprasad, School of Materials Science and Engineering
Abstract
Aromatic polymers, such as poly(ethylene terephthalate) (PET), polystyrene (PS), and polysulfone (PSU), are widely used because of the robust properties imparted by the aromatic units within their backbones. Unfortunately, these same properties prevent degradation and render their recycling and upcycling exceptionally difficult. Chemical post-polymerization modification (PPM) can produce materials with specialized properties, enabling their use in high-value industries. Unfortunately, traditional methods have limited substrate scope and can degrade the polymer chains. Gold-catalyzed direct C–H bond activation has recently been demonstrated as a method of modifying the aromatic rings of polymers, enabling the attachment of a wide range of pendants. The goal of this thesis is to determine how PPM changes the bulk and interfacial properties during and after synthesis. Aim 1 tests the bulk, determining the effect of pendant size, degree of functionalization, polarity, and intermolecular interactions on the thermomechanical and morphological properties. Aim 2 examines interfaces, testing the hypothesis that alkyl pendant length governs domain size, and mechanical properties in PET/polyethylene (PE) blends by controlling migration and anchoring. Aim 3 concludes with synthesis, determining whether catalyst structure and milling conditions allow PET chains to be functionalized with minimal degradation. Combined, these three aims will elucidate how PPM can be used to change the properties of aromatic polymers and their blends.
]]>
Ph.D. Thesis Defense Announcement
Sequencing-enabled Multi-Scale MicrobialProfiling for Environmental Bioprocesses: Portable Quantification,Microdiversity Characterization, and Epigenetic Early-Warning
By
Kaiqin Bian
Advisor:
Dr. Ameet Pinto
Committee Members:
Dr. Xing Xie (CEE)
Dr. Katy Graham (CEE)
Dr. Andrea Busch (Great Lakes Water Authority)
Dr. Raul Gonzalez (H2O Molecular)
Date and Time: Thursday, November 5th, 2026, 2:30 PM EDT
Location: Price Gilbert 4222
Virtual Attendance: Live Link
Teams Meeting ID: 293 127 571 504 326
will propose a doctoral thesis entitled,
Application of Chloroaluminate-Based Electrolytes in Lithium and Beyond-Lithium-Ion Batteries
On
Friday, November 6 at 9:30-11:30am
Closed
Abstract
The rapid expansion of electric mobility, grid-scale energy storage, and portable electronics necessitates next-generation rechargeable batteries that fundamentally transcend the safety risks, critical-mineral supply vulnerabilities, and high manufacturing costs of incumbent technologies. Conventional lithium-ion batteries rely on volatile, flammable organic carbonate electrolytes and cobalt/nickel-rich cathodes, presenting severe thermal-runaway hazards and geographic supply-chain dependencies. Consequently, developing earth-abundant, low-cost chemistries is essential for scalable energy storage. Among emerging candidate electrolyte platforms, chloroaluminate systems offer distinctive advantages spanning multiple battery chemistries. Enabled by facile coordination chemistry of chloroaluminates, these electrolytes can be tailored into ambient-temperature ionic liquids, deep eutectic analogues, or low-melting-point inorganic matrices.
This dissertation proposal outlines the investigation and application of chloroaluminate electrolytes across lithium and beyond-Li systems in two primary parts:
Part 1: Melt-Infiltrated Chloroaluminate Oxychloride Solid Electrolytes for Li and Na Batteries.
Conventional solid electrolytes suffer from poor solid–solid interfacial contact and processing constraints that limit active material mass loading. The first part proposes a novel chloroaluminate-based oxychloride inorganic solid electrolyte for both Li and Na systems. Preliminary findings demonstrate high room-temperature ionic conductivity and a modest thermal transition, allowing the electrolyte to behave as a low-viscosity liquid when heated to achieve complete, conformal melt-infiltration into tortuous composite electrodes, solidifying upon cooling into an intimate solid-state conducting network. The proposed research will evaluate the electrochemical performance, scalability, and interfacial chemo mechanics of this system to combine the manufacturing simplicity of liquid cells with the safety and mechanical stability of solid-state architectures.
Part 2: Na–Al Hybrid-Ion Batteries.
To transition toward abundant, low-cost battery platforms, the second part proposes a hybrid Na–Al battery chemistry coupling an aluminum metal anode with a dual-ion intercalating open-framework cathode operating in a ternary chloroaluminate ionic liquid electrolyte. The proposed work will systematically investigate the co-intercalation mechanisms of Na and Al species into the cathode host, assess the structural stability of the cathode during extended cycling, and determine the influence of salt concentration in the ionic liquid on ionic conductivity, transport properties, and overall cell performance.
Committee
]]>
(Advisor: Prof. Faisal Alamgir)
will defend a doctoral thesis entitled,
Revealing Anion Contributions in Lithium-Ion Batteries using Inner-Shell X-Ray Spectroscopy
On
Friday, November 6, 2026
12 pm - 2 pm ET
Callaway Manufacturing Research Building – Auditorium 101
or
Virtually via Microsoft Teams:
Meeting ID: 267 067 734 536 012
Passcode: hn9uZ366
Committee:
Prof. Faisal Alamgir – School of Material Science and Engineering (advisor)
Prof. Preet Singh – School of Material Science and Engineering
Prof. Matthew McDowell – School of Material Science and Engineering
Prof. Gleb Yushin – School of Material Science and Engineering
Prof. Yang Shao-Horn - Department of Material Science and Engineering at MIT
Dr. Jagjit Nanda - SLAC National Accelerator Laboratory
Abstract
The chemical state of elements in a lithium-ion battery (LIB) electrode provides valuable insights into the redox reactions occurring at the electrode site during cycling. While previous research has extensively investigated the role of cations in these reactions, the contribution of anions remains relatively unexplored. This is primarily due to the limited tools available to directly study the chemical state of anions. Inner-shell x-ray spectroscopy (ISXS) is well-suited for this purpose due to its element selectivity and ability of X-rays to probe electrode materials during in-situ or operando studies. Among these methods, x-ray absorption spectroscopy (XAS) examines the process of generating electron holes, offering insights into the electronic structure and local atomic arrangements of specific elements. Complementarily, x-ray emission spectroscopy (XES) studies the quenching of generated electron holes, providing sensitivity to chemical state, spin state, and ligand environment, making it an essential companion to XAS.
In this thesis, I will investigate the mechanism behind oxygen activity (redox) and their effects on the physical property of layered LiCoO2 (LCO), LiNixMnyCozO2 (NMCs, x + y + z = 1), and polyanion-based LiFePO4 (LFP) cathode material for LIBs. First, the onset of oxygen reactions in LCO and NMC cathodes is identified using a combination of several ISXS methods, followed by an investigation into the resulting changes in their structure and physical properties. Next, the formation of metastable phase during the charge process of LFP is qualitatively studied using linear combination analysis of iron K-edge XAS. The role of anions, specifically phosphorus and oxygen, in LFP is then investigated during charging using phosphorus K emissions and oxygen K-edge absorption, respectively. Finally, Kβ XES is explored as a probe for monitoring the chemical state of transition metals and their covalency with oxygen during LIB cycling, with its results benchmarked against those obtained using XAS.
]]>
School of Psychological and Brain Sciences – Ph.D. Dissertation Defense Meeting
Date: Thursday, November 12th, 2026
Time: 4:00 pm - 6:00 pm
Teams link: Join here
Dissertation Chair/Advisor:
Phillip L. Ackerman, Ph.D. (Georgia Institute of Technology)
Dissertation Committee Members:
Chris Wiese, Ph.D. (Georgia Institute of Technology)
Sashank Varma, Ph.D. (Georgia Institute of Technology)
Richard Catrambone, Ph.D. (Georgia Institute of Technology)
Chris Dede, Ph.D. (Harvard University)
Title: Creative Problem-Solving with Generative AI: An Examination of Aptitude-Treatment Interactions
Abstract: Organizations urge their employees to use generative artificial intelligence (AI) despite uncertainty about how it interacts with individual differences to influence job performance. A key individual difference linked to performance is cognitive ability; however, generative AI creates the potential for hybrid intelligence—human cognition augmented by AI—that may alter ability-performance relationships. To explore this possibility, I adopted an aptitude-treatment interaction (ATI) approach to test whether generative AI acts as a treatment that modifies the influence of cognitive and motivational aptitudes on creative problem-solving (CPS) performance. A sample of 111 undergraduate students participated in a three-part study that consisted of: 1) a survey measuring motivational traits and readiness to leverage generative AI; 2) cognitive ability tests; and 3) working on CPS tasks without and then with access to ChatGPT. For each task, participants read a scenario and prepared a short PowerPoint presentation of their solution. The results revealed three findings. First, solutions developed in the experimental condition (Internet plus ChatGPT) were rated higher than those in the control condition (Internet-use only), d = .72. Pilot data suggest the CPS tasks are similar in difficulty; however, because the tasks differed across conditions, these ratings cannot support claims that performance improved. Second, while none of the motivational aptitudes predicted performance in either condition, cognitive ability positively predicted performance in the control condition, B = .25, but not in the experimental condition, B = .07. Instead, generative AI readiness predicted performance in the experimental condition, B = .21, though the relationship was small. To better understand what differentiated high from low performance in the experimental condition, I reviewed screen recordings of the top and bottom performers to see how they used ChatGPT. High performers distinguished themselves by correctly anticipating that ChatGPT could adequately complete the task on its own. Those participants typed the full prompt into ChatGPT and then reallocated cognitive resources away from on-task activities (e.g., information gathering, ideation, reasoning) and put them toward self-regulatory activities that ChatGPT could not replace (e.g., managing time, modifying output, designing the presentation, ensuring it aligned with task expectations). This strategy is likely not optimal across all situations. Rather than viewing performance as a two-way aptitude-treatment interaction, performance may be better described as a three-way interaction that includes task complexity. That said, findings from the present study suggest that for relatively straightforward tasks requiring only general knowledge, generative AI may act as an equalizer, disproportionately supporting lower-ability individuals.
]]>
Thesis Title: “From Lipid Markers to Triboelectric Ion Sources: Mass Spectrometry Strategies For Monitoring Mesenchymal Stroma Cells and Extending Ionization”
Wednesday, November 4th, 2026, at 2:00 PM
Location: George Tower 0266
Or join via Teams: https://teams.microsoft.com/meet/2586090283616644?p=XUw56131tkAT0iQ9pR
Committee Members:
Prof. Facundo Fernández (Advisor) - School of Chemistry and Biochemistry
Prof. Neha Garg - School of Chemistry and Biochemistry
Prof. Ronghu Wu - School of Chemistry and Biochemistry
Prof. Craig Forest - School of Mechanical Engineering
Prof. Ahmet Coskun – Department of Biomedical Engineering
]]>Committee:
Dr. Van Hentenryck, Advisor
Dr. Molzahn, Co-Advisor
Dr. Meliopoulos, Chair
Dr. Crozier
]]>Committee:
Dr. Fabian Monrose, ECE, Chair, Advisor
Dr. Manos Antonakakis, ECE
Dr. Roberto Perdisci, UGA
Dr. Paul Pearce, UC Irvine
Dr. Angelos Keromytis, ECE
]]>ADVANCING GLOBAL ECO-HYDROLOGICAL MONITORING: AN INTEGRATED PHYSICAL MODELING FRAMEWORK FOR MULTI-SENSOR REMOTE SENSING
SEB 122
2026-11-19 2:00 PM EST
Rafael L. Bras
Passive microwave remote sensing offers extensive coverage of terrestrial water dynamics, yet reliable retrieval of hydrological variables remains challenging due to the complex interactions among soil, vegetation, and snow within the observed radiometric signal. This dissertation investigates an integrated physical modeling framework grounded in the Maximum Entropy Production (MEP) principle to improve the interpretation of microwave observations across heterogeneous terrestrial environments. Building upon radiative transfer theory and thermodynamically constrained energy partitioning, the MEP framework is examined for the simultaneous retrieval of soil moisture (SM) and vegetation water content (VWC). Evaluations across diverse ecosystems demonstrate strong agreement between retrieved soil moisture and ground observations, with correlations ranging from 0.74 to 0.91 at three SMAP core validation sites. Extensions of the framework reveal seasonal relationships between satellite-derived VWC and live fuel moisture in the western United States, while applications to the 2023–2024 Amazon drought identify widespread soil moisture deficits accompanied by spatially heterogeneous vegetation responses. A further extension to a two-layer snow–soil system demonstrates the potential for snow water equivalent (SWE) estimation using multi-frequency microwave observations, subject to prescribed snow properties and calibrated parameters. Together, these findings establish a physically interpretable foundation for investigating terrestrial water states through a common radiative-transfer and thermodynamic perspective. By demonstrating both the applicability and limitations of MEP-based retrievals, this dissertation contributes toward more reliable and integrated multi-sensor monitoring of terrestrial hydrological dynamics.
]]>
will propose a doctoral thesis entitled,
Spatial Gradient Architectures in Soft Elastomeric Materials with Expanded Functionality
On
Tuesday, November 17 at 1:00 p.m.
Molecular Science and Engineering Building (MoSE) Room 3201A
and
Virtually via Zoom
Abstract
Soft elastomeric materials, from elastomeric nanocomposites to soft bioelectronic devices, are conventionally designed with a uniform composition and geometry. This homogeneity imposes similar properties across the materials, limiting their ability to accommodate spatially varying material properties and functional demands and creating abrupt property mismatches at interfaces. Spatial gradient architectures, in which composition or geometry varies gradually and deliberately across a material, offer a strategy to distribute local properties according to these demands without introducing additional material components.
This work investigates spatial gradients in composition and geometry as independent design variables through two systems. In a six-layer laminated cellulose nanocrystal (CNC)/waterborne polyurethane (WPU) elastomer, the MXene content increases from 0 to 2 wt% at a fixed CNC/WPU ratio. At the same average loading as a uniform film, the compositional gradient film shows approximately 50% higher tensile strength while maintaining a comparable strain at break, along with a further 15% reduction in heat capacity change in the soft-segment region and a higher storage modulus, consistent with spatially heterogeneous polymer confinement induced by the compositional gradient. In a liquid metal (LM)/styrene-ethylene-butylene-styrene (SEBS) connector for ultrathin skin electrodes, the thickness increases from 0.5 to 10 μm over 5 cm. The connector follows skin topography, remains conductive to approximately 1100% strain and through 5,000 bending cycles, and gives an electrocardiography (ECG) signal-to-noise ratio of 20.1 dB during bending, compared with 13.7 dB for the corresponding commercial tape-based connector system. The thickness gradient is designed to combine the conformability of the thin region with the mechanical robustness of the thick region, providing a gradual transition that reduces mechanical mismatch across the device interface.
This work will provide a fundamental understanding of how spatial gradients in composition and geometry govern local material response and overall performance in soft elastomeric materials through studies of gradient architecture, local property distribution, and interfacial mechanics. The proposed research will further extend spatial gradient design to multifunctional material systems by integrating with different types of fillers and enantiomeric polymers to tailor mechanical, thermal, optical, and electrical responses.
Committee Members
Prof. Vladimir Tsukruk (Advisor), MSE
Prof. Meisha Shofner, MSE
Prof. Seung Soon Jang, MSE
Prof. Jerry Qi, ME
Prof. Todd Sulchek, ME
]]>Date: November 4th
Time: 9.00am (EST)
Room: 006
Zoom link: https://gatech.zoom.us/j/91925737499
Committee members:
Dr. Haomin Zhou (Advisor)
Dr. Peng Chen (Co-advisor)
Dr. Sung Ha Kang
Dr. Luca Dieci
Dr. Shu Liu
Abstract:
Finding a path between two probability distributions is a recurring problem in optimal transport, generative modeling, population dynamics, and control. A broad class of these problems can be formulated through an action functional defined over curves of probability distributions. The corresponding first-order optimality conditions lead to a coupled Hamilton–Jacobi and continuity system, which can be written as a boundary-value Hamiltonian flow in density space.
We take three approaches for learning and optimizing them: Hamiltonian Rectification (HR), Neural Multiple Shooting (NMS), and Parametric Density Path Optimization (PDPO). In the talk, we are going to focus on Neural Multiple Shooting.
The central idea of NMS is to reformulate the boundary-value problem in density space as a multiple-shooting problem in particle space. Since the particle dynamics are determined by the prescribed Hamiltonian system once the momentum is specified, the unknowns reduce to momentum maps that initialize a sequence of Hamiltonian initial-value problems. The resulting hybrid method combines neural approximation with classical multiple shooting: neural networks determine the momentum at the beginning of each shooting segment, while an ODE solver propagates the corresponding trajectories according to the prescribed dynamics.
]]>
Date: 2026-11-04
Time: 1pm-3pm
Location / Meeting Link: Clough Undergraduate Learning Commons, room 127; https://emory.zoom.us/j/91475318334
Committee Members:
Annabelle Singer, PhD (Advisor) Levi Wood, PhD Ming-Fai Fong, PhD Joseph Manns, PhD Jessica Bolton, PhD
Title: Microglial Regulation of Hippocampal Neural Activity During Spatial Learning
Abstract:
Spatial learning depends on coordinated activity in the hippocampus, a brain region essential for learning and memory. During brief bursts of activity called sharp-wave ripples, hippocampal neurons reactivate representations of past experiences, supporting memory consolidation. Microglia, the brain’s resident immune cells, interact with neurons and influence network activity, yet whether they contribute to the hippocampal changes required for new spatial learning remains unknown. Establishing their role in healthy animals will provide a foundation for understanding memory impairment in Alzheimer’s disease, where both microglial function and hippocampal activity are altered. This project will determine how microglia contribute to behavioral, network, and molecular changes during spatial learning in healthy adult mice. I hypothesize that microglia are engaged during learning and are necessary for the coordinated neuronal activity and gene-expression changes that support acquisition of new spatial information. Preliminary findings suggest increased expression of CD68, a marker associated with microglial activation and phagocytic activity, in hippocampal area CA3 after exposure to new reward locations. Behavioral findings also suggest that systemic microglial depletion may impair new spatial learning. To test this hypothesis, mice will first learn reward locations on a familiar virtual-reality track and then receive a microglia-depleting treatment directly into the hippocampus or a vehicle control. Recordings from hippocampal areas CA1 and CA3 during familiar-track performance and subsequent novel-track learning will assess spatial representations, sharp-wave ripple properties, and reactivation of newly learned spatial information. Bulk RNA sequencing of the hippocampus will identify learning-associated gene-expression changes and determine how depletion alters neuronal plasticity programs and responses in other brain cell populations. Vehicle-treated mice remaining on the familiar track will provide a comparison for identifying molecular responses associated with new learning. I predict that microglial depletion will preserve familiar-task performance but impair novel-track learning, reduce ripple-associated reactivation, and attenuate learning-associated neuronal plasticity programs. By connecting cellular responses with hippocampal activity and behavior, this work will test whether microglia actively support spatial learning and identify candidate pathways for future investigation of memory dysfunction in Alzheimer’s disease
Advisor: Dr. Elizabeth Qian
Milestone: PhD Thesis Final Examination (Defense)
Degree Program: Aerospace Engineering
Title: Data-Driven Nonlinear Reduced Model Learning with Stability Guarantee and Large-Scale Adaptation
Abstract: High-fidelity numerical models of engineering systems have become increasingly accurate, but their computational and memory costs remain a barrier to many-query tasks such as design optimization, uncertainty quantification, and control. The field of scientific machine learning (SciML) aims to address these costs by combining data-driven learning with physical knowledge to construct accurate, efficient surrogate models. This thesis develops new SciML methods based on Operator Inference (OpInf), a method that learns low-dimensional surrogate models from trajectory data without access to the underlying simulation code by fitting the operators of a prescribed polynomial model through least-squares regression. Although this structure facilitates analysis, it does not guarantee that the learned model inherits the stability of the original system. Moreover, the batch formulation requires that all training data be loaded into memory, limiting its applicability to large-scale systems. This thesis addresses these two challenges: stability and scalability. For stability, we propose Energy-Preserving OpInf, which constrains the learned quadratic operator to conserve energy, yielding provably bounded reduced dynamics when the linear dynamics are dissipative. We further propose a complementary method called Lyapunov Function Inference, which uses trajectory data alone to approximate solutions of Zubov’s equation, a nonlinear partial differential equation whose solution characterizes the stability region of an asymptotically stable equilibrium. By minimizing the equation’s residual over polynomial candidates, the method constructs Lyapunov functions to certify stability and estimate this region. For scalability, we develop Streaming OpInf, a method that combines streaming singular value decomposition with recursive least-squares to learn reduced models from incoming data without storing the full dataset. An accompanying error analysis bounds errors in the learned operators in terms of the incrementally updated subspaces. Beyond benchmark problems, we demonstrate Streaming OpInf on three large-scale turbulent flows that present significant memory challenges for batch OpInf: a compressible magnetohydrodynamic flow, a turbulent channel flow, and an atmospheric boundary layer flow with nearly 24 million states. We achieve models that reproduce relevant flow statistics while reducing memory requirements by 60–90%. Together, these contributions advance SciML for applications requiring stable learned dynamics and memory-efficient model construction.
Date and time: 2026-11-12, 2:30 pm to 4:30 pm
Location: MK317
Committee:
Dr. Elizabeth Qian (advisor), School of Aerospace Engineering
Dr. Keegan Moore, School of Aerospace Engineering
Dr. Joseph Oefelein, School of Aerospace Engineering
Dr. Benjamin Peherstorfer, École Polytechnique Fédérale de Lausanne
Dr. Christopher G. Baker, Nvidia Corporation
BioE PhD Proposal Presentation
Wednesday, November 4th, 2026, at 10:00 am
Location: IBB Suddath Seminar Room (1128)
Zoom Link: https://gatech.zoom.us/j/4256237492?omn=96368139923
Advisor:
Dr. Todd Sulchek, George W. Woodruff School of Mechanical Engineering
Committee Members:
Dr. Alexander Alexeev, George W. Woodruff School of Mechanical Engineering
Dr. Costas Arvanitis, George W. Woodruff School of Mechanical Engineering
Dr. Edikan Ogunnaike, Wallace H. Coulter Department of Biomedical Engineering
Dr. Shuichi Takayama, Wallace H. Coulter Department of Biomedical Engineering
Mechanics-Driven Microsystems for Cell Therapy Manufacturing
Cell therapies have transformed the treatment of cancer and hold promises for other hard-to-treat diseases, yet their broader adoption remains limited by the complexity, cost, and variability of manufacturing processes. Unlike conventional therapeutics, living cells are sensitive to mechanical forces encountered during manufacturing, from deformation during fluid processing to interactions with engineered materials during expansion. This research aims to advance the understanding of cell mechanical responses and develop microsystems that harness and control mechanical interactions during cell therapy manufacturing. In Aim 1, I will investigate how deformation rate and strain influence cell relaxation and damage responses, advancing fundamental understandings that inform the design of cell-processing systems. In Aim 2, I will develop a microfluidic workflow to label non-phagocytic cells with perfluorocarbon ultrasound contrast agents to enable non-invasive, real-time in vivo tracking at single cell level. A system of microfluidic devices will be used. First a microfluidic mixer will be used to spontaneous emulsification to form nanodroplets. Second, the resulting nanodroplets will then be directed to a microfluidic mechanoporation chip for delivery into T-cells. In Aim 3, I will develop more cell-like antigen-presenting particles with controlled size and stiffness using a feedback-controlled microfluidic platform to produce hydrogel particles. The feedback-control platform uses real-time independent assessments of particle size and stiffness to modify the flow of input materials to achieve a dramatic reduction in property drift. As a result, synthetic antigen-presenting hydrogel particles with cell-like properties will be created and used to investigate how mechanical properties can be tuned to enhance T-cell expansion and promote desirable memory and stem-like phenotypes. Together, these studies will bridge fundamental cell mechanics and engineering applications, laying the foundation for more controlled and reliable cell therapy manufacturing.
]]>Mechanics-Driven Microsystems for Cell Therapy Manufacturing
]]>Committee:
Dr. Callie Hao, ECE, Chair, Advisor
Dr. Divya Mahajan, ECE
Dr. Celine Lin, CS
Dr. Shuyi Pei, Samsung
Dr. Yiran Chen, Duke
Dr. Yuan Xie, Hong Kong UST
]]>]]>
School of Psychological and Brain Sciences – Ph.D. Dissertation Defense Meeting
Date: Tue, Nov 10th, 2026
Time: 10:00 AM - 11:00 AM EST
Location: https://gatech.zoom.us/my/kaixue
Dissertation Committee Chair/Advisor:
Dobromir Rahnev, Ph.D. (Georgia Tech)
Dissertation Committee Members:
Tansu Celikel, Ph.D. (Georgia Tech)
Sashank Varma, Ph.D. (Georgia Tech)
Mengyao Li, Ph.D. (Georgia Tech)
Megan Peters, Ph.D. (University College London)
Title: The Computation of Confidence in Perceptual Decision-Making
Abstract: This dissertation investigates the computational mechanisms underlying confidence in perceptual decision-making through three related studies. Study 1 uses transcranial magnetic stimulation (TMS) to establish when confidence is computed, revealing that the dorsolateral prefrontal cortex (DLPFC) contributes to confidence during a broad temporal window overlapping with decision-making. Study 2 tests competing computational theories — Bayesian Confidence Hypothesis (BCH) versus Confidence in Raw Evidence Space (CRES) — in two-choice tasks, finding consistent support for evidence-based rather than probability-based confidence computation. Study 3 extends this investigation to multi-alternative decisions using a novel dot numerosity paradigm, demonstrating that confidence tracks the difference between the two highest evidence values rather than choice probability. Together, these studies advance a mechanistic framework in which confidence emerges in parallel with decisions and operates on sensory evidence rather than probability estimates.
]]>