I will discuss recent work on high-dimensional inference from nonlinear and quantized observations, focusing on generalized linear measurement models. I will describe information-theoretic limits on recovery and simple algorithms that approach these limits under structural assumptions such as sparsity. I will also discuss extensions to models with latent structure, including mixtures of generalized linear models and quantitative group testing, highlighting how aggregation and heterogeneity reshape fundamental trade-offs in sample complexity and identifiability.
Short Bio: Neha Sangwan is currently a visiting researcher at the Tata Institute of Fundamental Research (TIFR) and previously held a postdoctoral position at the University of California San Diego. She received her PhD from TIFR, and her research focuses on information-theoretic foundations of high-dimensional inference, sequential decision-making in quantum systems, and reliable communication under adversarial uncertainty.
