Priyanka Sinha

I am a Departmental Postdoctoral Fellow in Mathematics at Virginia Tech. My research explores how optimization, probability, and learning theory can be combined to solve large-scale problems that arise in imaging, signal processing, and data analysis.

A current focus is on distributed dynamic inverse problems, where I design algorithms for reconstructing time-varying tomographic images from partial measurements. This work blends operator theory, convex optimization, and high-performance computing to address the computational limits of centralized solvers.

I also study fairness, retention, and efficiency in federated learning. By framing aggregation as a repeated game over clustered probability measures, I investigate how correlated equilibria can guide the trade-off between accuracy, communication, and fairness across clients.

Earlier in my career I developed methods for adversarial learning and watermarking in deep neural networks, adaptive meta-learning for communication systems, and statistical signal processing for detection and tracking of aerial vehicles. These experiences shape my interest in principled algorithms that remain robust in practical settings.

Across all projects, I aim to build theory-driven methods that are computationally efficient and well suited for real-world applications.