Batch-sequential design and heteroskedastic surrogate modeling for delta smelt conservation
From MaRDI portal
Publication:2154181
Abstract: Delta smelt is an endangered fish species in the San Francisco estuary that have shown an overall population decline over the past 30 years. Researchers have developed a stochastic, agent-based simulator to virtualize the system, with the goal of understanding the relative contribution of natural and anthropogenic factors suggested as playing a role in their decline. However, the input configuration space is high-dimensional, running the simulator is time-consuming, and its noisy outputs change nonlinearly in both mean and variance. Getting enough runs to effectively learn input--output dynamics requires both a nimble modeling strategy and parallel supercomputer evaluation. Recent advances in heteroskedastic Gaussian process (HetGP) surrogate modeling helps, but little is known about how to appropriately plan experiments for highly distributed simulator evaluation. We propose a batch sequential design scheme, generalizing one-at-a-time variance-based active learning for HetGP surrogates, as a means of keeping multi-core cluster nodes fully engaged with expensive runs. Our acquisition strategy is carefully engineered to favor selection of replicates which boost statistical and computational efficiencies when training surrogates to isolate signal in high noise regions. Design and modeling performance is illustrated on a range of toy examples before embarking on a large-scale smelt simulation campaign and downstream high-fidelity input sensitivity analysis.
Recommendations
- Sequential design for computer experiments with a flexible Bayesian additive model
- Design and analysis of simulation experiments
- A novel hybrid sequential design strategy for global surrogate modeling of computer experiments
- Bayesian-validated computer-simulation surrogates for optimization and design: Error estimates and applications
- Optimal design for correlated processes with input-dependent noise
Cites work
- scientific article; zbMATH DE number 1529823 (Why is no real title available?)
- scientific article; zbMATH DE number 3799842 (Why is no real title available?)
- A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code
- A Limited Memory Algorithm for Bound Constrained Optimization
- An efficient algorithm for Elastic I‐optimal design of generalized linear models
- Analyzing stochastic computer models: a review with opportunities
- Batch sequential designs for computer experiments
- Bayesian calibration of computer models. (With discussion)
- Calibrating a stochastic, agent-based model using quantile-based emulation
- Computer experiment designs for accurate prediction
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Discrete Optimization via Simulation Using COMPASS
- Efficient global optimization of expensive black-box functions
- Evaluating Gaussian process metamodels and sequential designs for noisy level set estimation
- Exploratory designs for computational experiments
- Locally induced Gaussian processes for large-scale simulation experiments
- Microcolony and biofilm formation as a survival strategy for bacteria
- Microsimulation model calibration using incremental mixture approximate Bayesian computation
- Optimal predictive model selection.
- Practical Heteroscedastic Gaussian Process Modeling for Large Simulation Experiments
- Probabilistic Sensitivity Analysis of Complex Models: A Bayesian Approach
- Sequential Learning of Active Subspaces
- Stochastic kriging for simulation metamodeling
- Strictly Proper Scoring Rules, Prediction, and Estimation
- The design and analysis of computer experiments
This page was built for publication: Batch-sequential design and heteroskedastic surrogate modeling for delta smelt conservation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2154181)
