model-calibration-evaluation
Evaluating model calibration methods for sensitivity analysis, uncertainty analysis, optimisation, and Bayesian inference
See config.yaml for the ground-truth simulation parameters.
The following model calibration methods have been evaluated.
- Approximate Bayesian Computation - Sequential Monte Carlo
- Bayesian Optimisation
- Bayesian Optimisation for Likelihood-Free Inference
- Differential Evolution Adaptive Metropolis
- Experimental Design via Gaussian Process Emulation
- Flow Matching Posterior Estimation
- Tree-structured Parzen Estimator
- Polynomial Chaos Expansion
- Polynomial Chaos Kriging
- Sparse Axis-Aligned Subspace Bayesian Optimization
- Shuffled Complex Evolution Algorithm Uncertainty Analysis
- Sequential Neural Posterior Estimation
- Sobol Sensitivity Analysis
- Truncated Marginal Neural Ratio Estimation