Abstract | ||
---|---|---|
Particle filtering - perhaps more properly named Sequential Monte Carlo - approaches have a strong potential for signal and image processing applications. A problem of great practical significance in this field, which remains largely unsolved as of today, is the esti- mation of fixed model parameters based on the output of sequential simulations. In this contribution, we investigate maximum likelihood esti- mation approaches based either on gradient or EM (Expectation- Maximization) techniques and show that several recently proposed methods share the common feature of requiring the approximation of the expectation of a sum functional of the hidden states, condi- tionally on all the available observations. Considering this general task, we discuss empirical results concerning the influence of the number of particles and sample size. We also propose a robustifi- cation of the basic particle estimator which is based on forgetting ideas. |
Year | Venue | Keywords |
---|---|---|
2005 | EUSIPCO | expectation-maximisation algorithm,gradient methods,particle filtering (numerical methods),expectation-maximization technique,fixed model parameter estimation,forgetting ideas,gradient technique,maximum likelihood parameter estimation,particle filter,particle number,robust basic particle estimator,sample size,sequential monte carlo method,sequential simulations |
Field | DocType | ISBN |
Monte Carlo method,Mathematical optimization,Robustification,Particle filter,Algorithm,Image processing,Estimation theory,Maximum likelihood sequence estimation,Hidden Markov model,Mathematics,Estimator | Conference | 978-160-4238-21-1 |
Citations | PageRank | References |
7 | 1.59 | 3 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
O. Cappe | 1 | 2112 | 207.95 |
Eric Moulines | 2 | 3248 | 337.64 |