Research

My research studies sequential decision making and optimization in multi agent systems where information, objectives, or data distributions are heterogeneous and evolve over time. 

Research interests

• Distributed and decentralized optimization
  
  

• Federated learning under heterogeneity
  
  

• Online convex optimization and regret based dynamics
  
  

• Large scale inverse problems and tomographic reconstruction
  
  

• Communication efficient and adaptive learning algorithms

Selected research themes

Distributed dynamic inverse problems
I design decentralized consensus based optimization frameworks for inverse problems such as total variation regularized tomography, where different nodes observe different parts of the underlying signal. My work introduces adaptive, coordinate wise consensus mechanisms that encode heterogeneous observability and reliability across nodes, together with regret based updates that learn effective communication patterns over time.

Dynamic coreset selection for large scale learning
I develop adaptive coreset selection methods that evolve during training, using continuous relaxations and greedy optimization techniques to preserve gradient accuracy while significantly reducing data usage. These methods aim to balance computational efficiency and statistical performance in large scale machine learning.

Federated learning with no regret dynamics
I study federated learning through a game theoretic lens, separating server side aggregation from client side optimization. By using external regret minimization and adaptive clustering, my work improves fairness, retention, and convergence behavior in heterogeneous federated systems while reducing communication overhead.

Publications

Journal articles

P. Sinha and M. Kang, Correlated Equilibrium for Enhanced Retention in Federated Learning with Optimal Aggregation Communication Trade off, in preparation for Journal of Machine Learning Research, 2026.

P. Sinha, N. Alberti, S. Kang, M. Matthews, and G. Matthews, Dynamic Coreset Selection of Large Scale Machine Learning, in preparation for Journal of Machine Learning Research, 2026.

P. Sinha and I. Guvenc, Impact of Antenna Pattern on TOA Based 3D UAV Localization Using a Terrestrial Sensor Network, IEEE Transactions on Vehicular Technology, 2022.

P. Sinha, M. M. U. Chowdhury, I. Guvenc, D. W. Matolak, and K. Namuduri, Wireless Connectivity and Localization for Advanced Air Mobility Services, IEEE Aerospace and Electronic Systems Magazine, 2022.

P. Sinha, I. Guvenc, and M. C. Gursoy, Fundamental Limits on Detection of UAVs by Existing Terrestrial RF Networks, IEEE Open Journal of the Communications Society, vol. 2, 2021, pp. 2111 to 2130.

Conference publications

P. Sinha, J. Kibilda, and W. Saad, On the Tradeoff Between Heterogeneity and Communication Complexity in Federated Learning, Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, California, 2023.

P. Sinha, H. Krim, and I. Guvenc, Neural Network Based Tracking of Maneuvering Unmanned Aerial Vehicles, Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, California, 2022.

M. M. U. Chowdhury, P. Sinha, and I. Guvenc, Handover Count Based Velocity Estimation of Cellular Connected UAVs, International Workshop on Signal Processing Advances in Wireless Communications, Atlanta, Georgia, 2020.

P. Sinha, Y. Yapici, I. Guvenc, E. Turgut, and M. C. Gursoy, RSS Based Detection of Drones in the Presence of RF Interferers, IEEE Consumer Communications and Networking Conference, Las Vegas, Nevada, 2020.

P. Sinha, Y. Yapici, and I. Guvenc, Impact of 3D Antenna Radiation Patterns on TDOA Based Wireless Localization of UAVs, IEEE INFOCOM Workshops, Paris, France, 2019.

J. Chen, Y. Zhou, D. Raye, W. Khawaja, P. Sinha, and I. Guvenc, Impact of 3D UWB Antenna Radiation Pattern on Air to Ground Drone Connectivity, IEEE Vehicular Technology Conference, Chicago, Illinois, 2018.

Contact

Email: prnk.snh@gmail.com, yaana@vt.edu 

Google Scholar: https://scholar.google.com/priyankasinha

GitHub: https://github.com/prsinha1/ 

For academic correspondence, collaboration inquiries, or speaking invitations, email is the best way to reach me.