1 / 31

Yosuke Morishima

Yosuke Morishima. Pattern Recognition and Machine Learning. Chapter 13 : Sequential Data. Contents in latter part . Linear Dynamical Systems What is different from HMM? Kalman filter Its strength and limitation Particle Filter Its simple algorithm and broad application

lamar
Télécharger la présentation

Yosuke Morishima

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Yosuke Morishima Pattern Recognition and Machine Learning Chapter 13: Sequential Data

  2. Contents in latter part Linear Dynamical Systems What is different from HMM? Kalman filter Its strength and limitation Particle Filter Its simple algorithm and broad application Application in Neuroscience

  3. Dynamical Systems State space model of Dynamical systems Now latent variable is continuous rather than discrete State equation Observation equation

  4. Dynamical Systems State space model of Dynamical systems Now latent variable is continuous rather than discrete or

  5. Linear Dynamical Systems Special case of dynamical systems Gaussian assumption on distribution Where z: Latent variable x: Observation A: Dynamics matrix C: Emission matrix w, v, u: Noise

  6. Linear Dynamical Systems Bayesian formof LDS Since LDS is linear Gaussian model, joint distribution over all latent and observed variables is simply Gaussian.

  7. Kalman Filter Kalman filter does exact inference in LDS in which all latent and observed variables are Gaussian (incl. multivariate Gaussian). Kalman filter handles multiple dimensions in a single set of calculations. Kalman filter has two distinct phases: Predict and Update

  8. Application of Kalman Filter Tracking moving object Blue: True position Green: Measurement Red: Post. estimate

  9. Two phases in Kalman Filter Predict Prediction of state estimate and estimate covariance Update Update of state estimate and estimate covariance with Kaman gain

  10. Estimation of Parameter in LDS Distribution of Zn-1 is used as a prior for estimation of Zn Predict Update Red: Blue: Red: Green: Blue:

  11. Derivation of Kalman Filter We use sum-product algorithm for efficient inference of latent variables. LDS is continuous case of HMM (sum -> integer) where

  12. Sum-product algorithm where Solve… where (Kalman Gain Matrix)

  13. What we have estimated? Predict: Update:

  14. What we have estimated? Prediction error Predict: Update: Predicted mean of Zn Predicted mean of Xn Observed Xn

  15. Parameter Estimation with EM algorithm E step: Given , run the inference algorithm to determine distribution of latent variable M step: Given , maximize the complete-data log likelihood function wrtq where Where

  16. Limitation of Kalman Filter Due to assumption of Gaussian distribution in KF, KF can not estimate well in nonlinear/non-Gaussian problem. One simple extension is mixture of Gaussians In mixture of K Gaussians, is mixture of K Gaussians, and will comprise mixture of Kn Gaussians. -> Computationally intractable

  17. Limitation of Kalman Filter To resolve nonlinear dynamical system problem, other methods are developed. Extended Kalman filter: Gaussian approximation by linearizing around the mean of predicted distribution Particle filter: Resampling method, see later Switching state-space model: continuous type of switching HMM

  18. Particle Filter In nonlinear/non-Gaussian dynamical systems, it is hard to estimate posterior distribution by KF. Apply the sampling-importance-resampling (SIR) to obtain a sequential Monte Carlo algorithm, particle filter. Advantages of Particle Filter Simple algorithm -> Easy to implement Good estimation in nonlinear/non-Gaussian problem

  19. How to Represent Distribution Original distribution (mixture of Gaussian) Gaussian approximation Approximation by PF (distribution of particle) Approximation by PF (histogram)

  20. Where’s a landmine? Use metal detector to find a landmine (orange star).

  21. Where’s a landmine? Random survey in the field (red circle). The filter draws a number of randomly distributed estimates, called particles. All particles are given the same likelihood

  22. Where’s a landmine? Get response from each point (strength: size). Assign a likelihood to each particle such that the particular particle can explain the measurement.

  23. Where’s a landmine? Decide the place to survey in next step. Scale the weight of particles to select the particle for resampling

  24. Where’s a landmine? Intensive survey of possible place Draw random particles based upon their likelihood (Resampling). High likelihood -> more particle; low likelihood -> less particle All particle have equal likelihood again.

  25. Operation of Particle Filter

  26. Algorithm of Particle Filter Sample representation of the posterior distribution p(zn|Xn) expressed as a samples {z(l)n} with corresponding weights {w(l)n}. Draw samples from mixture distribution Use new observation xn+1 to evaluate the corresponding weights . where where

  27. Limitation of Particle Filter In high dimensional models, enormous particles are required to approximate posterior distribution. Repeated resampling cause degeneracy of algorithm. All but one of the importance weights are close to zero Avoided by proper choice of resampling method

  28. Application to Neuroeconomics Estimation of Q value (latent variable) in reward tracking task Prew(L)-Prew (R) Samejima, Science 05

  29. Model framework a: learning rate g: discount factor b: temperature parameter Q: action value a: action r: reward

  30. Estimation of Q

  31. Demonstration Estimation of Q value in reward seeking task Code is provided by KazuSamejima Comparison of performance btw with/without resampling. Code is provided by KlaasEnno Stephan

More Related