data assimilation and kernel
play

Data Assimilation and Kernel Reconstruction for Nonlocal Field - PowerPoint PPT Presentation

Data Assimilation and Kernel Reconstruction for Nonlocal Field Dynamics Roland Potthast DWD & University of Reading and Jehan Alswaihli University of Reading ISDA Kobe 2019 Contents 1. Introduction and Amari Equation 2. Neural State


  1. Data Assimilation and Kernel Reconstruction for Nonlocal Field Dynamics Roland Potthast DWD & University of Reading and Jehan Alswaihli University of Reading ISDA Kobe 2019

  2. Contents 1. Introduction and Amari Equation 2. Neural State Estimation 3. Neural Kernel Problem (= Deep Learning) 4. Integrated State and Kernel estimation

  3. How to use neural field models in reality?

  4. Amari / Cowan-Wilson Equation

  5. Amari Equation Solvability: Fixed Point Theorem

  6. Amari Equation Example: Oscillator

  7. Amari Equation Kernel for Oscillator

  8. Contents 1. Introduction and Amari Equation 2. Neural State Estimation 3. Neural Kernel Problem (= Deep Learning) 4. Integrated State and Kernel estimation

  9. • Consider some Pulse or Signal • Measured at some given points (tiny electrodes) • Or as integrated values (large electrodes)

  10. Classical State Estimation

  11. Covariance Matrix B

  12. Singular Values of H for large electrode case

  13. State Estimation Results

  14. Contents 1. Introduction and Amari Equation 2. Neural State Estimation 3. Neural Kernel Problem (= Deep Learning) 4. Integrated State and Kernel estimation

  15. A deep learning algorithm = inverse problem solution:

  16. A deep learning algorithm = inverse problem solution:

  17. A deep learning algorithm = inverse problem solution:

  18. Solution with different Regularization Parameters all with strong input noise (>10%)

  19. Contents 1. Introduction and Amari Equation 2. Neural State Estimation 3. Neural Kernel Problem (= Deep Learning) 4. Integrated State and Kernel estimation

  20. Estimation and Reconstruction

  21. Original Pulse Measurements Estimate Simulation after Rec

  22. Est-Rec-Iteration

  23. Convergence Result (Alswaihli and P.) • The Transport Map is bounded • The Estimator is convergent and bounded • The Reconstruction is convergent and bounded As a consequence, the iteration is convergent. For noisy data you need a stopping rule.

  24. Original Pulse and Simulated Pulse from reconstructed Kernel Iteration 2 Iteration 1 Iteration 5 Iteration 4

  25. After 20 time steps, Iterations 1-5 After 25 time steps, Iterations 1-5 Original Pulse and Iterations from reconstructed Kernel

  26. Simulated Pulse Original Pulse from learned / reconstructed Kernel (no noise)

  27. Many Thanks!

Recommend


More recommend