| 65 | | BDM is a collection of methods for selected tasks of Bayesian decision-making, such as estimation, filtering and control.<h2><a class="anchor" name="Approach"> |
| 66 | | Approach</a></h2> |
| 67 | | The toolbox is designed using two design criteria: <ul> |
| | 65 | BDM is a collection of methods for selected tasks of Bayesian decision-making, such as estimation, filtering and control.<h2><a class="anchor" name="fea"> |
| | 66 | Features</a></h2> |
| | 67 | At present the following algorithms are implemented:<ul> |
| | 68 | <li><b>Bayesian</b> <b>filtering</b> : Kalman filter, EKF, patricle filter (PF),<ul> |
| | 69 | <li>these can be combined mutualy together in mode demanding schemes, see marginalized particle filter MPF</li></ul> |
| | 70 | </li></ul> |
| | 71 | <p> |
| | 72 | <ul> |
| | 73 | <li><b>Classification</b> using mixtures of exponential famiuly models (MixEF),</li><li><b>Density</b> <b>estimation</b> : using mixtures (MixEF), density composition (merger)</li></ul> |
| | 74 | <h2><a class="anchor" name="down"> |
| | 75 | Download and Use</a></h2> |
| | 76 | The library is available under GPL, see installation instructions on page <a class="el" href="install.html">How to install BDM</a><p> |
| | 77 | Precompiled Mex files for use within Matlab are available <a class="el" href="mexfiles.html">How to write and use mex files for Matlab</a><h2><a class="anchor" name="app"> |
| | 78 | Design Approach</a></h2> |
| | 79 | The toolbox is designed using <b>object-oriented</b> approach with two design criteria: <ul> |
| 71 | | Hence, each mathematical object such as probability density is represented by one software object. The resulting algorithms are then implemented as operations on these objects. In cases when more efficient solution can be achived when this structure is not respected, a parallel implementation is created and clearly marked as specific.<h2><a class="anchor" name="Methods"> |
| 72 | | Methods</a></h2> |
| 73 | | At present we support the following methods: <ul> |
| 74 | | <li>Estimation: ARX models </li> |
| 75 | | <li>Filtering: Kalman filtering, Extended Kalman filters (EKF), Particle filters (PF), Marginalized Particle filters (MPF). </li> |
| 76 | | <li>Desing: None.</li> |
| 77 | | </ul> |
| 78 | | <h2><a class="anchor" name="Implementation"> |
| | 83 | Hence, each mathematical object such as probability density is represented by one software object. The resulting algorithms are then implemented as operations on these objects. In cases when more efficient solution can be achived when this structure is not respected, a parallel implementation is created and clearly marked as specific.<p> |
| | 84 | OpenMP is used to achive efficient implementation on parallel architectures.<h2><a class="anchor" name="impl"> |
