Estimation of time series by Maximum Mean Discrepancy
By: Pierre Alquier, Jean-David Fermanian, Benjamin Poignard
Potential Business Impact:
Finds best math models for tricky data.
We define two minimum distance estimators for dependent data by minimizing some approximated Maximum Mean Discrepancy distances between the true empirical distribution of observations and their assumed (parametric) model distribution. When the latter one is intractable, it is approximated by simulation, allowing to accommodate most dynamic processes with latent variables. We derive the non-asymptotic and the large sample properties of our estimators in the context of absolutely regular/beta-mixing random elements. Our simulation experiments illustrate the robustness of our procedures to model misspecification, particularly in comparison with alternative standard estimation methods.
Similar Papers
Maximum Mean Discrepancy with Unequal Sample Sizes via Generalized U-Statistics
Machine Learning (Stat)
Keeps all your data for better tests.
A Martingale Kernel Two-Sample Test
Methodology
Finds differences between groups faster.
Ensemble Transport Filter via Optimized Maximum Mean Discrepancy
Machine Learning (Stat)
Improves computer predictions by learning from past data.