Solving Constrained CASH Problems with ADMM
This addresses the need for more flexible automated ML configuration for practitioners, though it appears incremental as it builds on existing CASH and ADMM methods.
The paper tackled the problem of automated machine learning pipeline configuration (CASH) not handling black-box constraints like fairness or robustness, and presented an approach using ADMM to decompose CASH into smaller problems and incorporate such constraints.
The CASH problem has been widely studied in the context of automated configurations of machine learning (ML) pipelines and various solvers and toolkits are available. However, CASH solvers do not directly handle black-box constraints such as fairness, robustness or other domain-specific custom constraints. We present our recent approach [Liu, et al., 2020] that leverages the ADMM optimization framework to decompose CASH into multiple small problems and demonstrate how ADMM facilitates incorporation of black-box constraints.