deletions | additions       diff --git a/subsection_Visualization_of_result_Once__.tex b/subsection_Visualization_of_result_Once__.tex  index c28b9af..1f011b4 100644  --- a/subsection_Visualization_of_result_Once__.tex  +++ b/subsection_Visualization_of_result_Once__.tex 
...
\subsection{Visualization of result}  Once the distribution probability distributions  and joint distributions are estimated, the MI-based distance between two all couple of  trajectories $d_{MI}(T_{i},T_{j})$  can be calculated using the forms $(7)$ and $(8)$. $(8)$, gathered in a distance matrix.   \[M = \left( \begin{array}{ccc}  d_{MI}(T_{1},T_{1}) & \cdots & d_{MI}(T_{1},T_{n}) \\  \vdots & & \vdots \\  d_{MI}(T_{1},T_{n}) & \cdots & d_{MI}(T_{n},T_{n}) \end{array} \right).\]  And then the clustering of the whole set of trajectories can be calculated using the clustering algorithm of \cite{NIPS2008_3478}. Then come the thorny issue of interpretation of the given clustering. To this purpose, as a first approach we animate the trajectories belonging to the same cluster.