wasserstein distributionally robust kalman filtering
play

Wasserstein Distributionally Robust Kalman Filtering (1) (1) (1) - PowerPoint PPT Presentation

Wasserstein Distributionally Robust Kalman Filtering (1) (1) (1) Soroosh Shafieezadeh-Abadeh, Viet Anh Nguyen, Daniel Kuhn (2) and Peyman Mohajerin Esfahani (1) Risk Analytics and Optimization Chair, EPFL (2) Delft University of Technology


  1. Wasserstein Distributionally Robust Kalman Filtering (1) (1) (1) Soroosh Shafieezadeh-Abadeh, Viet Anh Nguyen, Daniel Kuhn (2) and Peyman Mohajerin Esfahani (1) Risk Analytics and Optimization Chair, EPFL (2) Delft University of Technology Poster: AB #14

  2. Minimum mean square error estimation Zero-sum game against nature Wasserstein distance 1

  3. Optimal estimator Least favorable prior Structural results Result 1. Result 2. Frank-Wolfe algorithm Result 3. Analytically solvable oracle subproblem Result 4. Guaranteed convergence speed Image source: Jaggi, ICML (2013) 2

  4. Classical Kalman filter state estimate observation system update predict nominal prior: 3

  5. Distributionally robust Kalman filter state estimate observation system update predict Least favorable prior: nominal prior: solve SDP via Frank-Wolfe algorithm 4

  6. Numerical results MMSE estimation Frank-Wolfe algorithm Kalman filtering Robustness reduces regret Empirical convergence speed Wasserstein filter displays: Lowest steady-state error Fastest convergence POSTER: AB #14 5

Recommend


More recommend