Evaluating Gaussian process metamodels and sequential designs for noisy level set estimation
From MaRDI portal
Publication:2058770
Abstract: We consider the problem of learning the level set for which a noisy black-box function exceeds a given threshold. To efficiently reconstruct the level set, we investigate Gaussian process (GP) metamodels. Our focus is on strongly stochastic samplers, in particular with heavy-tailed simulation noise and low signal-to-noise ratio. To guard against noise misspecification, we assess the performance of three variants: (i) GPs with Student- observations; (ii) Student- processes (TPs); and (iii) classification GPs modeling the sign of the response. In conjunction with these metamodels, we analyze several acquisition functions for guiding the sequential experimental designs, extending existing stepwise uncertainty reduction criteria to the stochastic contour-finding context. This also motivates our development of (approximate) updating formulas to efficiently compute such acquisition functions. Our schemes are benchmarked by using a variety of synthetic experiments in 1--6 dimensions. We also consider an application of level set estimation for determining the optimal exercise policy of Bermudan options in finance.
Recommendations
- Sequential design of experiments for estimating percentiles of black-box functions
- Gaussian Process Regression on Nested Spaces
- A supermartingale approach to Gaussian process based sequential design of experiments
- Sequential design of computer experiments for the estimation of a probability of failure
- Parallel Gaussian process surrogate Bayesian inference with noisy likelihood evaluations
Cites work
- A Knowledge-Gradient Policy for Sequential Information Collection
- Bayesian Subset Simulation
- Comparison of Kriging-based algorithms for simulation optimization with heterogeneous noise
- Efficient global optimization of expensive black-box functions
- Extended t-process regression models
- Gaussian processes for machine learning (GPML) toolbox
- Gaussian processes for machine learning.
- GPstuff: Bayesian modeling with Gaussian processes
- scientific article; zbMATH DE number 3656961 (Why is no real title available?)
- Information-Theoretic Regret Bounds for Gaussian Process Optimization in the Bandit Setting
- Practical Heteroscedastic Gaussian Process Modeling for Large Simulation Experiments
- Quantifying uncertainties on excursion sets under a Gaussian random field prior
- Regression Level Set Estimation Via Cost-Sensitive Classification
- Robust Gaussian process regression with a Student-t likelihood
- Sequential design for optimal stopping problems
- Sequential design for ranking response surfaces
- Sequential design of computer experiments for the estimation of a probability of failure
- Stochastic kriging for simulation metamodeling
- The design and analysis of computer experiments
- Valuing American options by simulation: a simple least-squares approach
Cited in
(4)- Analyzing stochastic computer models: a review with opportunities
- Batch-sequential design and heteroskedastic surrogate modeling for delta smelt conservation
- Active learning for level set estimation under input uncertainty and its extensions
- Finite-dimensional approximation of Gaussian processes with linear inequality constraints and noisy observations
This page was built for publication: Evaluating Gaussian process metamodels and sequential designs for noisy level set estimation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2058770)
