diogo/yap-6.3
Archived
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

PFL manual: rework the parameter learning section

Browse Source
This commit is contained in:
Tiago Gomes 2013-01-12 12:30:34 +00:00
parent 90614d3594
commit 83ccb31665
1 changed files with 41 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
packages/CLPBN/pfl.tex
View File

@ -400,10 +400,50 @@ This option allows to print a textual representation of the factor graph.
\section{Parameter Learning} \section{Parameter Learning}
PFL is capable to learn the parameters for bayesian networks, through an implementation of the expectation-maximization algorithm. PFL is capable to learn the parameters for bayesian networks, through an implementation of the expectation-maximization algorithm.
Inside the \texttt{learning} directory from the examples directory, one can find some examples of how learning works in PFL. Next we show an example of parameter learning for the sprinkler network.
\begin{pflcode}
:- [sprinkler.pfl].
:- use_module(library(clpbn/learning/em)).
data(t, t, t, t).
data(_, t, _, t).
data(t, t, f, f).
data(t, t, f, t).
data(t, _, _, t).
data(t, f, t, t).
data(t, t, f, t).
data(t, _, f, f).
data(t, t, f, f).
data(f, f, t, t).
main :-
findall(X, scan_data(X), L),
em(L, 0.01, 10, CPTs, LogLik),
writeln(LogLik:CPTs).
scan_data([cloudy(C), sprinkler(S), rain(R), wet_grass(W)]) :-
data(C, S, R, W).
\end{pflcode}
Parameter learning is done by calling the \texttt{em/5} predicate. Its arguments are the following.
\texttt{em(+Data, +MaxError, +MaxIters, -CPTs, -LogLik)}
Where,
\begin{itemize}
\item \texttt{Data} is a list of samples for the distribution that we want to estimate. Each sample is a list of either observed random variables or unobserved random variables (denoted when its state value is not instantiated).
\item \texttt{MaxError} is the maximum error allowed before stopping the EM loop.
\item \texttt{MaxIters} is the maximum number of iterations for the EM loop.
\item \texttt{CPTs} is a list with the estimated conditional probability tables.
\item \texttt{LogLik} is the log-likelihood.
\end{itemize}
It is possible to choose the solver that will be used for the inference part during parameter learning with the \texttt{set\_em\_solver/1} predicate (defaults to \texttt{hve}). At the moment, only the following solvers support parameter learning: \texttt{ve}, \texttt{hve}, \texttt{bdd}, \texttt{bp} and \texttt{cbp}. It is possible to choose the solver that will be used for the inference part during parameter learning with the \texttt{set\_em\_solver/1} predicate (defaults to \texttt{hve}). At the moment, only the following solvers support parameter learning: \texttt{ve}, \texttt{hve}, \texttt{bdd}, \texttt{bp} and \texttt{cbp}.
Inside the \texttt{learning} directory from the examples directory, one can find more examples of parameter learning.