Changeset 984

Show
Ignore:
Timestamp:
05/25/10 23:04:57 (14 years ago)
Author:
smidl
Message:

doc

Location:
library/doc/local
Files:
1 added
1 modified

Legend:

Unmodified
Added
Removed

  • library/doc/local/mainpage.dox

    r302 r984  
    77 
    88BDM is a collection of methods for selected tasks of Bayesian decision-making, such as estimation, filtering and control. 
     9It is implemneted in C++ with available interfaces to matlab (called bdmtoolbox) and Python (preliminary). 
    910 
    1011\section fea Features 
    1112At present the following algorithms are implemented: 
    12 - \b Bayesian \b filtering : Kalman filter, EKF, patricle filter (PF), 
     13 - \b Bayesian \b filtering : Kalman filter, EKF, patricle filter (PF), 
    1314    - these can be combined mutualy together in mode demanding schemes, see marginalized particle filter MPF 
    1415 
    15 - \b Classification using mixtures of exponential famiuly models (MixEF), 
    16 - \b Density \b estimation : using mixtures (MixEF), density composition (merger) 
     16 - \b Classification using mixtures of exponential famiuly models (MixEF), 
     17 - \b Density \b estimation : using mixtures (MixEF), density composition (merger) 
     18 - \b LQG \b control : so far only for ARX and Kalman filters 
    1719 
    1820\section down Download and Use 
    1921The library is available under GPL, see installation instructions on page \ref install 
    2022 
    21 Precompiled Mex files for use within Matlab are available \ref mexfiles 
     23It is split into library and applications. One of the applications is toolbox for matlab, which can be downloaded in binary form for win32, see \ref install. 
    2224 
    23 \section app Design Approach 
    24 The toolbox is designed using \b object-oriented approach with two design criteria: 
    25 \li relation to mathematics, 
    26 \li efficient evaluation, 
    27 while the first one is more important than the latter. 
     25\section app Publications 
    2826 
    29 Hence, each mathematical object such as probability density is  
    30 represented by one software object. The resulting algorithms are  
    31 then implemented as operations on these objects. In cases when  
    32 more efficient solution can be achived when this structure is not respected,  
    33 a parallel implementation is created and clearly marked as specific.  
    34  
    35 OpenMP is used to achive efficient implementation on parallel architectures. 
     27Publications that were made with the toolbox are: 
     28 - \b Software papers: 
     29 - \b Distributed \b identification : 
     30 - \b Baysian \b filtering : 
     31  
     32The status of replicability of the published experiments is available in \ref published. 
    3633 
    3734\section impl Implementation 
    3835 
    39  BDM is build on top of \c IT++ which wraps numerically efficient operations of linear algebra into easy to use C++ classes. Thanks to this excellent library, writing of numerical algorithms is as easy as in Matlab but we gain significant advantages: 
    40  
    41 \li computational speed comparable to built-in Matlab function, and surpassing interpreted Matlab in order of magnitudes, 
    42 \li native support for object-oriented programming, 
    43 \li support for templates which is often more appropriate than object-oriented programming, 
    44 \li cross-platform compatibility. 
    45  
    46  
    47 */ 
     36 BDM is build on top of \c IT++ which wraps numerically efficient operations of linear algebra into easy to use C++ classes with Matlab-like syntax. 
     37 */