Title: Comparative survey on non linear filtering methods : the quantization and the particle filtering approaches Afef SELLAMI
1Comparative survey on non linear filtering
methods thequantization and the particle
filtering approachesAfef SELLAMI
2Overview
- Introduction
- Bayes filters
- Quantization based filters
- Zero order scheme
- First order schemes
- Particle filters
- Sequential importance
- sampling (SIS) filter
- Sampling-Importance
- Resampling(SIR) filter
- Comparison of two approaches
- Summary
3Non linear filter estimators
- Quantization based filters
- Zero order scheme
- First order schemes
- Particle filtering algorithms
- Sequential importance
- sampling (SIS) filter
- Sampling-Importance
- Resampling(SIR) filter
4Overview
- Introduction
- Bayes filters
- Quantization based filters
- Zero order scheme
- First order schemes
- Particle filters
- Sequential importance
- sampling (SIS) filter
- Sampling-Importance
- Resampling(SIR) filter
- Comparison of two approaches
- Summary
5Bayes Filter
- Bayesian approach We attempt to construct the
pnf of the state given all measurements. - Prediction
- Correction
6Bayes Filter
- One step transition bayes filter equation
- By introducint the operaters ,
sequential definition of the unnormalized filter
pn - Forward Expression
7Overview
- Introduction
- Bayes filters
- Quantization based filters
- Zero order scheme
- First order schemes
- Particle filters
- Sequential importance
- sampling (SIS) filter
- Sampling-Importance
- Resampling(SIR) filter
- Comparison of two approaches
- Summary
8Quantization based filters
- Zero order scheme
- First order schemes
- One step recursive first order scheme
- Two step recursive first order scheme
9Zero order scheme
- Quantization
- Sequential definition of the unnormalized filter
pn - Forward Expression
10Zero order scheme
11Recalling Taylor Series
- Let's call our point x0 and let's define a new
variable that simply measures how far we are
from x0 call the variable h x x0. - Taylor Series formula
- First Order Approximation
12First order schemes
- Introduce first order schemes to improve the
convergence rate of the zero order schemes. - Rewriting the sequential definition by mimicking
some first order Taylor expansion - Two schemes based on the different approximation
by - One step recursive scheme based on a recursive
definition of the differential term estimator. - Two step recursive scheme based on an
integration by part transformation of conditional
expectation derivative.
13One step recursive scheme
- The recursive definition of the differential term
estimator - Forward Expression
14(No Transcript)
15Two step recursive scheme
- An integration by part formula
- where
- where
16(No Transcript)
17Comparisons of convergence rate
- Zero order scheme
- First order schemes
- One step recursive first order scheme
- Two step recursive first order scheme
18Overview
- Introduction
- Bayes filters
- Quantization based filters
- Zero order scheme
- First order schemes
- Particle filters
- Sequential importance
- sampling (SIS) filter
- Sampling-Importance
- Resampling(SIR) filter
- Comparison of two approaches
- Summary
19Particle filtering
- Consists of two basic elements
- Monte Carlo integration
- Importance sampling
20Importance sampling
Proposal distribution easy to sample from
Original distribution hard to sample from, easy
to evaluate
Importance weights
21Sequential importance sampling (SIS) filter
- we want samples from
- and make the following importance sampling
identifications
Proposal distribution
Distribution from which we want to sample
22SIS Filter Algorithm
23Sampling-Importance Resampling(SIR)
- Problems of SIS
- Weight Degeneration
- Solution ? RESAMPLING
- Resampling eliminates samples with low importance
weights and multiply samples with high importance
weights - Replicate particles when the effective number of
particles is below a threshold
24Sampling-Importance Resampling(SIR)
Prediction
Resampling
Update
Sensor model
x
25Overview
- Introduction
- Bayes filters
- Quantization based filters
- Zero order scheme
- First order schemes
- Particle filters
- Sequential importance
- sampling (SIS) filter
- Sampling-Importance
- Resampling(SIR) filter
- Comparison of two approaches
- Summary
26Elements for a comparison
- Complexity
- Numerical performances in three state models
- Kalman filter (KF)
- Canonical stochastic volatility model (SVM)
- Explicit non linear filter
27Complexity comparison
Zero order scheme C0N2
One step recursive first order scheme C1N2d3
Two step recursive first order scheme C2N2d
SIS particle filter C3N
SIR particle filter C4N
28Numerical performances
- Three models chosen to make up the benchmark.
- Kalman filter (KF)
- Canonical stochastic volatility model (SVM)
- Explicit non linear filter
29Kalman filter (KF)
- Both signal and observation equations are linear
with Gaussian independent noises. - Gaussian process which parameters (the two first
moments) can be computed sequentially by a
deterministic algorithm (KF)
30Canonical stochastic volatility model (SVM)
- The time discretization of a continuous diffusion
model. - State Model
31Explicit non linear filter
- A non linear non Gaussian state equation
- Serial Gaussian distributions SG()
- State Model
32Numerical performance Results
- Convergence tests
- three test functions
- Kalman filter d1
33Numerical performance Results Convergence rate
improvement
- ltRegression slopes on the log-log scale
representation (d3)gt
34Numerical performance Results
- Stochastic volatility model
ltParticle filter for large particle sizes (N
10000) and quantization filter approximations for
SVM as a function of the quantizer sizegt
35Numerical performance Results
- Non linear explicit filter
ltExplicit filter estimators as function of grid
sizes gt
36Conclusions
- Particle methods do not suffer from dimension
dependency when considering their theoretical
convergence rate, whereas quantization based
methods do depend on the dimension of the state
space. - Considering the theoretical convergence results,
quantization methods are still competitive till
dimension 2 for zero order schemes and till
dimension 4 for first order ones. - Quantization methods need smaller grid sizes than
Monte Carlo methods to attain convergence regions