High-dimensional Black-box Optimization Under Uncertainty

Speaker: 

Hadis Anahideh, Research Assistant Professor of Mechanical and Industrial Engineering, UIC

Description: 

limited informative data remains the primary challenge for the optimization expensive complex systems. Learning from limited data and finding the set of variables that optimizes an expected output arise practically everywhere from molecular structure design for drug discovery to a deep neural network tuning. In such situations, the underlying function is complex, yet unknown, a large number of variables are involved, though not all of them are important, and the interaction between the variables is significant. On the other hand, it is usually expensive to collect more data and the outcome is under uncertainty. Unfortunately, despite being real-world challenges, exiting work have not addressed these jointly. In this talk, I will present a new surrogate optimization paradigm to address these primary concerns. We propose an algorithm for the global optimization of computationally expensive black-box functions designing a flexible, noninterpolating, and parsimonious surrogate model using a partitioning technique. Furthermore, we develop a smart replication approach based on hypothesis testing to overcome the uncertainties associated with the black-box system.

Event Topic:

Computational Mathematics & Statistics