In this talk, I will present my work in designing and analyzing principled methodologies for emerging learning and estimation tasks in resource-constrained, heterogeneous, collaborative networks, with a vision towards socially responsible use of such distributed systems. I will particularly discuss my contributions in developing efficient algorithms with mathematical guarantees for 1) communication-efficient collaborative learning in heterogeneous networks, 2) Information sharing and observation selection in collaborative sensing systems, and 3) leveraging latent parsimony in data-scarce scenarios for effective function approximation and inference. Finally, I will conclude this talk by discussing my future research on addressing the challenges of having resilient, secure, and effective collaborative learning systems that are socially responsible and beneficial.

Bio