New Techniques for Algorithm Portfolio Design
This work solves the algorithm portfolio design problem for AI and optimization domains, but it is incremental as it builds on prior scheduling work.
The paper tackles the problem of algorithm portfolio design by addressing both scheduling and machine learning aspects simultaneously, resulting in improved performance for state-of-the-art algorithms in Boolean satisfiability, zero-one integer programming, and A.I. planning.
We present and evaluate new techniques for designing algorithm portfolios. In our view, the problem has both a scheduling aspect and a machine learning aspect. Prior work has largely addressed one of the two aspects in isolation. Building on recent work on the scheduling aspect of the problem, we present a technique that addresses both aspects simultaneously and has attractive theoretical guarantees. Experimentally, we show that this technique can be used to improve the performance of state-of-the-art algorithms for Boolean satisfiability, zero-one integer programming, and A.I. planning.