Amin Karbasi

Papers

Total Citations

H-Index

About

Amin Karbasi is a leading researcher in machine learning and artificial intelligence, whose work fundamentally advances how we make decisions under uncertainty. His core research areas include Bayesian active learning, adaptive submodularity, and safe optimization. Karbasi’s major contributions lie in developing theoretically grounded frameworks for sequential decision-making. He pioneered the use of submodular surrogates for the value of information, solving the intractable problem of optimizing information-gathering strategies with near-optimal guarantees—a breakthrough with over 43 citations. His work on "Near Optimal Bayesian Active Learning for Decision Making" (43 citations) provides a rigorous foundation for selecting tests to reduce hypothesis uncertainty, while his research on "Adaptivity in Adaptive Submodularity" (2019) addresses the central challenge of interactive policy design. More recently, Karbasi has tackled the critical problem of safe learning under uncertain objectives and constraints (2020), developing methods for non-convex optimization in safety-critical domains like robotics and medical procedures. His contributions have been recognized through prestigious awards, including an NSF CAREER Award and multiple best paper nominations. Karbasi’s research elegantly bridges theoretical guarantees with practical algorithms, making him a pivotal figure in modern decision theory and AI safety.