Research

Shrinkage-Gated Uncertainty Estimation: Combining Prior Beliefs and Density Estimates (Nevena Gligić, Arya Farahi)

(Ongoing)
Goal: Enhance AI system reliability under distributional shifts by developing an architecture that leverages the known structure of the in-distribution data through a density estimation model.
Impact: Fraud detection, medical diagnosis, autonomous driving, disaster response management.

BPSD: Unsupervised Bayesian Probabilistic Signal Detection in Noisy Environments (Nevena Gligić, Arya Farahi)

(Ongoing)
Goal: Develop a framework for unsupervised probabilistic signal detection under low signal-to-noise ratio (SNR) conditions without assumptions about the signal distribution.
Impact: Gravitational waves detection, galaxies detection, medical imaging, radar systems.

AI-Driven Utility Monitoring and Anomaly Detection for City-Operated Buildings (Arya Farahi, Matt Kammer-Kerwick, Nevena Gligić, Vineet Burugu)

(Ongoing)
Goal: Build an AI infrastructure for energy forecasting and anomaly detection within an interactive dashboard.
Impact: Improved operational efficiency, reduced waste, lower emissions, cut costs.