Upcoming Doctoral Dissertations

"Traditional graph theory has been widely adopted to address a broad range of graph-structured data problems. Recent advances in artificial intelligence have successfully integrated neural network techniques into network analysis. In particular, Graph Neural Networks (GNNs) have demonstrated remarkable effectiveness in learning node representations and identifying clustering patterns in graph-structured data." Read more

"In the acute respiratory distress syndrome (ARDS), a form of acute lung injury, inflammation increases the permeability of pulmonary capillaries. As a result, proteinaceous liquid leaks from the capillaries into the smallest pulmonary airspaces, the alveoli. The condition is known as alveolar edema. Additionally, surface tension at the interface of a thin liquid layer that lines the alveoli is elevated. Edema and elevated surface tension impair oxygenation." Read more

"Artificial intelligence has achieved remarkable success in medical image analysis, yet translating these advances into clinical practice remains challenging. In real-world healthcare applications, medical AI models often face limited and imbalanced datasets, imperfect supervision, small and heterogeneous anatomical structures, and the risk of producing clinically inaccurate outputs." Read more

"Point cloud data is often acquired in biomechanical and industrial applications to synthesize custom-fit components that conform to existing surface geometries. Converting these scans into usable CAD models remains largely manual, and automated alternatives fit surfaces with a single loss function applied across the entire geometry, yielding representations whose smoothness and fit quality are neither controlled nor quantified." Read more

"Automated cell characterization is increasingly important in biomedical imaging because modern
experiments generate large and complex image datasets that are difficult to analyze manually with
consistency and throughput. This dissertation develops artificial-intelligence-based tools for converting
biomedical images into biologically meaningful measurements across multiple applications. " Read more

"Electron transport in semiconductors can be significantly influenced by external magnetic fields, giving rise
to magnetotransport phenomena such as the quantum Hall effect, spin–orbit coupling, and magnetoresistance.
These effects have attracted considerable interest due to their potential applications in next-generation
optoelectronic and spintronic devices. Among semiconducting systems, two-dimensional (2D) materials are
particularly promising because of their tunable electronic and magnetic properties, strong spin-orbit coupling,
and diverse crystal symmetries." Read more

"Code clones are common in software systems because developers frequently reuse code through copy-and-paste. Although this practice can improve development efficiency, it also increases maintenance effort by requiring duplicated code to be updated consistently. Extract Method refactoring is a widely used technique for removing duplicated code while preserving program behavior." Read more

"Communication skills are essential for individuals to thrive in today’s fast-paced and socially demanding society. The ability to interact effectively in both informal and formal settings is a key component of personal and professional development. Virtual Reality (VR)–based simulations have emerged as a promising medium for training such soft skills by enabling immersive, repeatable, and controlled social interactions. This work is motivated by the observation that an individual’s social competence is closely linked to the level of confidence they exhibit in social settings." Read more

"Adaptive neural networks (AdNNs) have enjoyed wide adoption as an alternative to traditional deep neural networks (DNNs) due to their ability to dynamically adjust computation during inference. By leveraging optimization mechanisms such as early-exits and other adaptive inference strategies, AdNNs can reduce latency, computational cost, memory requirements, and energy consumption of traditional DNNs without significantly sacrificing their accuracy, making them particularly attractive for deployment in resource-constrained environments such as mobile devices, IoT systems, and real-time low-latency applications." Read more

"Effective thermal management is essential to the safety, performance, and service life of lithium-ion batteries, particularly under the rapid charge-discharge rates encountered in electric-vehicle applications. Phase change materials (PCMs) provide an attractive passive cooling solution because of their high latent heat capacity; however, their inherently low thermal conductivity limits heat dissipation and motivates enhancement strategies such as conductive nanoparticles or embedded metal fins." Read more

"This dissertation develops mathematical modeling and structure-aware design frameworks for two complex
physical systems whose behavior is controlled by geometry and field response. The first part studies electrohydrodynamic deformation of toroidal drops in electric fields. For a prescribed toroidal interface with elliptic cross section, an inverse-optimization framework is used to determine the permittivity ratio and electric capillary number that minimize the residual normal velocity" Read more

"The three essays examine how executive incentives, internal pay structures, and corporate governance shape public firms. They study CEO compensation across firm life cycles, executive pay dispersion under financing constraints, and changes in CEO relative pay and power following TCJA-induced tax windfalls." Read more