B0936
Title: Hybrid method for constrained Bayesian optimization
Authors: Neelesh Shankar Upadhye - Indian Institute of Technology Madras (India) [presenting]
Raju Chowdhury - Indian Institute of Technology Madras (India)
Abstract: A hybrid Bayesian optimization framework is proposed using an indicator function to solve constrained optimization problems efficiently. In the hybrid method, a hybrid acquisition function is derived to effectively guide the search that explores both the feasible and infeasible regions. The proposed method is demonstrated on five test problems, including two black-box problems, also a hyperparameter tuning problem to evaluate the performance of the hybrid method and compare it with the state-of-the-art methods available in the literature.
