A Nonparametric Approach to Noisy and Costly Optimization

This paper describes Pairwise Bisection: a nonparametric approach to optimizing a noisy function with few function evaluations. The algorithm uses nonparametric reasoning about simple geometric relationships to nd minima e ciently. Two factors often frustrate optimization: noise and cost. Output can contain signi cant quantities of noise or error, while time or money allows for only a handful of experiments. Pairwise bisection is used here to attempt to automate the process of robust and e cient experimentdesign. Real world functions also tend to violate traditional assumptions of continuousness and Gaussian noise. Since nonparametric statistics do not depend on these assumptions, this algorithm can optimize a wide variety of phenomena with fewer restrictions placed on noise. The algorithm's performance is compared to that of three competing algorithms, Amoeba, PMAX, and Q2 on several di erent test functions. Results on these functions indicate competitive performance and superio...

• Type: Conference
• Year: 2000
• Where: ICML
• Authors: Brigham S. Anderson, Andrew W. Moore, David Cohn
• Keywords: Gaussian Noise, ICML 2000, Machine Learning, Nonparametric Approach, Nonparametric Reasoning