Abstract | ||
---|---|---|
Computer experiments often require dense sweeps over input parameters to obtain a qualitative understanding of their response. Such sweeps can be prohibitively expensive, and are unnecessary in regions where the response is easy predicted; well-chosen designs could allow a mapping of the response with far fewer simulation runs. Thus, there is a need for computationally inexpensive surrogate models and an accompanying method for selecting small designs. We explore a general methodology for addressing this need that uses non-stationary Gaussian processes. Binary trees partition the input space to facilitate non-stationarity and a Bayesian interpretation provides an explicit measure of predictive uncertainty that can be used to guide sampling. Our methods are illustrated on several examples, including a motivating example involving computational fluid dynamics simulation of a NASA reentry vehicle. |
Year | DOI | Venue |
---|---|---|
2004 | 10.1145/1015330.1015367 | ICML |
Keywords | Field | DocType |
parameter space exploration,input space,binary tree,computational fluid dynamics simulation,bayesian interpretation,computationally inexpensive surrogate model,accompanying method,gaussian process tree,nasa reentry vehicle,computer experiment,input parameter,fewer simulation,gaussian process,parameter space | Computer experiment,Computer science,Binary tree,Gaussian process,Sampling (statistics),Artificial intelligence,Parameter space,Computational fluid dynamics,Partition (number theory),Machine learning,Bayesian probability | Conference |
ISBN | Citations | PageRank |
1-58113-838-5 | 24 | 3.70 |
References | Authors | |
7 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Robert Gramacy | 1 | 240 | 30.15 |
Herbert K. H. Lee | 2 | 209 | 34.39 |
William G. Macready | 3 | 161 | 39.07 |