Probabilistic function-on-function nonlinear autoregressive model for emulation and reliability analysis of stochastic dynamical systems
Researchers have introduced a new Function-on-Function Nonlinear AutoRegressive model with eXogenous inputs (F2NARX), which enhances predictive efficiency and accuracy for complex dynamical systems. This novel framework combines principal component analysis with Gaussian process regression, enabling probabilistic predictions through an unscented transform in an autoregressive manner. The F2NARX model significantly outperforms existing NARX models in both speed and precision, and its active learning capabilities allow for accurate estimation of first-passage failure probabilities with minimal training data. AI
IMPACT This new modeling technique could improve the efficiency and accuracy of simulations in engineering and scientific research.