AAS 390 Chapter Notes - Chapter 1: Performance Metric, Amazon Web Services, Computer-Aided Design
Document Summary
Daniel geschwender1, frank hutter2, lars kotthoff3, yuri malitsky3, 1university of nebraska-lincoln, 2university of freiburg, 3insight centre for data. Con guring algorithms automatically to achieve high performance is becoming increas- ingly relevant and important in many areas of academia and industry. Guration methods take a parameterized target algorithm, a performance metric and a set of example data, and aim to nd a parameter con guration that performs as well as possible on a given data set. Gga [1], irace [2], and smac [4] have achieved impressive performance improve- ments in a broad range of applications. However, these systems often require substantial computational resources to nd good con gurations. With the advent of cloud comput- ing, these resources are available readily and at moderate cost, offering the promise that these techniques can be applied even more widely. However, the use of cloud com- puting for algorithm con guration raises two challenges.