Worst-case quadratic loss bounds for prediction using linear functions and gradient descent
IEEE Transactions on Neural Networks
Randomness in a mathematical model for the transmission of respiratory syncytial virus (RSV)
Mathematics and Computers in Simulation
| Hi-index | 0.09 |
The objective of this article is to investigate the topics related to uncertainties in air quality modeling. A first point is the evaluation of uncertainties for model outputs: Monte Carlo methods and sensitivity analysis are powerful methods for assessing the impact of uncertainties due to model inputs. A second point is devoted to ensemble modeling with multi-models approaches. According to the wide spread in the model outputs, using a unique model, tuned to a small set of observational data, is not relevant in this field. On the basis of ensemble simulations, improved forecasts are given by appropriate algorithms to combine the set of models. The results applied to air quality modeling at continental scale with the Polyphemus system illustrate these methods. The first estimates of uncertainties in inverse modeling experiments are also proposed.