What is SPASM - Software for Probabilistic Analysis, Storage, and Manipulation?
SPASM is a templatised C++ library for the storage and manipulation of a variety of probabilistic representations.
Represenationtations currently considered are: Gaussian, Gaussian Mixtures, Parzen density Estimates, Particles and Discrete grids.
this library will include:
Other libraries for Bayesian filtering exist such as Bayes++
and BFL .
These libraries are focused more on filtering in general and
not the methods used for storage and manipulation of a variety
- Storage classes for PDF's and likelihoods
- Basic operators such as multiplication, division, convolution and addition.
- Conversion between common types (e.g. Gaussian)
- Information measures (such as entropy etc.)
- Distance measures
- Complexity Reduction (Gaussian Mixtures have a habit of growing after common operations)
This library has been motivated by robotics applications, such as feature tracking, SLAM and other forms of localisation. Numerous other applications exist that require probability representations, such a statistical learning.
The BOOST ublas library provides basic storage and manipulation of
vectors and matrices. It is intended that this library will fill this
role for Probability Density Functions (PDF) and Likelihood functions.
This library will make use of the BOOST ublas matrix library.
Visit the project page at http://sourceforge.net/projects/spasm/