Algorithm Configuration: Learning Policies for the Quick Termination of Poor Performers

• Published in: Learning and Intelligent Optimization Vol. 11353; pp. 220 - 224
• Main Authors: Karapetyan, Daniel, Parkes, Andrew J., Stützle, Thomas
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 01.01.2019; Springer International Publishing
• Series: Lecture Notes in Computer Science
• ISBN: 3030053474; 9783030053475
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-05348-2_20

One way to speed up the algorithm configuration task is to use short runs instead of long runs as much as possible, but without discarding the configurations that eventually do well on the long runs. We consider the problem of selecting the top performing configurations of Conditional Markov Chain Search (CMCS), a general algorithm schema that includes, for example, VNS. We investigate how the structure of performance on short tests links with those on long tests, showing that significant differences arise between test domains. We propose a “performance envelope” method to exploit the links; that learns when runs should be terminated, but that automatically adapts to the domain.