Turned into english
This commit is contained in:
parent
885b87a36e
commit
e3000067f2
2 changed files with 80 additions and 20 deletions
BIN
rapport.pdf
BIN
rapport.pdf
Binary file not shown.
100
rapport.tex
100
rapport.tex
|
|
@ -10,6 +10,8 @@
|
|||
\usepackage{epsfig} % pour gérer les images
|
||||
\usepackage{amsmath,amsthm} % très bon mode mathématique
|
||||
\usepackage{amsfonts,amssymb,bm, bbold}% permet la definition des ensembles
|
||||
\usepackage{algorithm2e} % pour les algorithmes
|
||||
\usepackage{algpseudocode} % pour les algorithmes
|
||||
\usepackage{float} % pour le placement des figure
|
||||
\usepackage{url} % pour une gestion efficace des url
|
||||
\usepackage{hyperref} % pour les hyperliens dans le document
|
||||
|
|
@ -34,9 +36,9 @@ Formule du point fixe pour la distribution de Bernoulli
|
|||
% Repasser à l'exponentielle pour la présentation du point fixe
|
||||
\begin{itemize}
|
||||
\item \textit{iid} :
|
||||
\[ \log(\bm{\tau}^{m,1}) = ~^{t}\log(\pi) + (\text{Mask}^{m} \odot A^{m})
|
||||
\[ \bm{\tau}^{m,1} = ~^{t}\pi + \exp[(\text{Mask}^{m} \odot A^{m})
|
||||
\bm{\tau}^{m,2} ~^{t}(\text{logit}(\alpha)) + \text{Mask}^{m}
|
||||
\bm{\tau}^{m,2} ~^{t}\log(\bm{1} - \alpha) \]
|
||||
\bm{\tau}^{m,2} ~^{t}\log(\bm{1} - \alpha)] \]
|
||||
\[ \log(\bm{\tau}^{m,2}) = ~^{t}\log(\rho) + ~^{t}(\text{Mask}^{m} \odot A^{m})
|
||||
\bm{\tau}^{m,1} \text{logit}(\alpha) + ~^{t}\text{Mask}^{m}
|
||||
\bm{\tau}^{m,1} \log(\bm{1} - \alpha) \]
|
||||
|
|
@ -52,45 +54,48 @@ Formule du point fixe pour la distribution de Bernoulli
|
|||
avec $\text{Mask}^{m}$ la matrice qui contient des $0$ si la valeur est un NA et
|
||||
des $1$ sinon.
|
||||
|
||||
\section{Calcul des pénalités}
|
||||
\section{M step of the algorithm}
|
||||
|
||||
\section{Computation of the variational bound}
|
||||
|
||||
\section{Penalties}
|
||||
|
||||
\paragraph*{\textit{iid-colBiSBM}}
|
||||
Dans le cas \textit{iid-colBiSBM} les pénalités sont à adapter de la manière
|
||||
suivante :
|
||||
For the \textit{iid-colBiSBM} the penalties were modified in the following way :
|
||||
|
||||
\begin{itemize}
|
||||
\item Pour les $\pi$ et les $\rho$ :
|
||||
\item For the $\pi$s and $\rho$s:
|
||||
\[\text{pen}_{\pi}(Q_1) = (Q_1 - 1)\log(\sum_{m=1}^{M}n_{r}^{(m)})\]
|
||||
\[\text{pen}_{\rho}(Q_2) = (Q_2 - 1)\log(\sum_{m=1}^{M}n_{c}^{(m)})\]
|
||||
\item Pour les $\alpha$ :
|
||||
\item For the $\alpha$s :
|
||||
\[\text{pen}_{\alpha}(Q_1, Q_2) = Q_1 \times Q_2 \log(N_M)\]
|
||||
avec
|
||||
\[ N_M = \sum_{m = 1}^{M} n_{r}^{(m)} \times n_{c}^{(m)} \]
|
||||
\end{itemize}
|
||||
Le BIC-L devient donc :
|
||||
\[ \text{BIC-L}(\bm{X},Q_1, Q_2) = \max_{\theta} \mathcal{J} (\mathcal{\hat{R}}, \bm{\theta})
|
||||
And thus the $BIC-L$ formula is now:
|
||||
\[ \text{$BIC-L$}(\bm{X},Q_1, Q_2) = \max_{\theta} \mathcal{J} (\mathcal{\hat{R}}, \bm{\theta})
|
||||
- \frac{1}{2} [\text{pen}_{\pi}(Q_1) + \text{pen}_{\rho}(Q_2) + \text{pen}_{\alpha}(Q_1, Q_2)]\]
|
||||
|
||||
\paragraph*{\textit{$\rho\pi$-colBiSBM}}
|
||||
Dans le cas \textit{$\rho\pi$-colBiSBM} les pénalités deviennent les suivantes :
|
||||
For the \textit{$\rho\pi$-colBiSBM} the penalties are the following:
|
||||
|
||||
\begin{itemize}
|
||||
\item Pour les pénalités dûes aux supports :
|
||||
\item The support penalties are:
|
||||
\[ \text{pen}_{S_1}(Q_1) = -2 \log p_{Q_1} (S_1) \]
|
||||
\[ \text{pen}_{S_2}(Q_2) = -2 \log p_{Q_2} (S_2) \]
|
||||
avec
|
||||
with
|
||||
\[ \log p_{Q_1}(S_1) = - M \log(Q_1) - \sum_{m=1}^{M} \log {Q_1 \choose Q_1^{(m)}} \]
|
||||
\[ \log p_{Q_2}(S_2) = - M \log(Q_2) - \sum_{m=1}^{M} \log {Q_2 \choose Q_2^{(m)}} \]
|
||||
\item Pour les pénalités dûes aux $\rho$ et $\pi$ :
|
||||
\item Penalties for the $\rho$s and $\pi$s:
|
||||
\[ \text{pen}_{\pi}(Q_1, S_1) = \sum_{m=1}^{M} (Q_{1}^{(m)} - 1) \log n_{r}^{(m)} \]
|
||||
\[ \text{pen}_{\rho}(Q_2, S_2) = \sum_{m=1}^{M} (Q_{2}^{(m)} - 1) \log n_{c}^{(m)} \]
|
||||
\item Pour les pénalités dûes aux $\alpha$ :
|
||||
\item Penalties for the $\alpha$s:
|
||||
\[ \text{pen}_{\alpha}(Q_1, Q_2, S_1, S_2) = (\sum_{q=1}^{Q_1} \sum_{r=1}^{Q_2} \mathbb{1}_{(S_1)'S_2 > 0}) \log (N_M) \]
|
||||
\end{itemize}
|
||||
Et alors le BIC-L est égal à :
|
||||
And the corresponding $BIC-L$ formula:
|
||||
\[
|
||||
\begin{aligned}
|
||||
\text{BIC-L}(\bm{X},Q_1, Q_2) =
|
||||
\text{$BIC-L$}(\bm{X},Q_1, Q_2) =
|
||||
\max_{S_1,S_2} [
|
||||
& \max_{\theta_{S_1,S_2} \in \Theta_{S_1,S_2}} \mathcal{J}(\mathcal{\hat{R}},\theta_{S_1,S_2})\\
|
||||
- \frac{1}{2} & (\text{pen}_{\pi}(Q_1, S_1) + \text{pen}_{\rho}(Q_2, S_2)\\
|
||||
|
|
@ -99,12 +104,67 @@ Et alors le BIC-L est égal à :
|
|||
\end{aligned}
|
||||
\]
|
||||
|
||||
\section{Exploration de l'espace latent}
|
||||
\subsection{Initialisation et appariemment des modèles}
|
||||
\section{Latent space exploration and model selection}
|
||||
In order to explorer the bi-dimensional latent space $(Q_1,Q_2)$
|
||||
we use the following strategies.
|
||||
|
||||
\subsection{Exploration gloutonne pour trouver une estimation du mode}
|
||||
\subsection{Model selection}
|
||||
In the following steps the model selection consists of using the $BIC-L$
|
||||
criterion to select the model. We choose among the proposed models the one that
|
||||
maximizes the $BIC-L$
|
||||
|
||||
\subsection{Fenêtre glissante pour mettre à jour les clusterings et les BIC-L}
|
||||
\subsection{Initialization and pairing of the models}
|
||||
First to combine the information from the $M$ networks we fit a collection model
|
||||
for each network at the two points $Q = (1, 2)$ and $Q = (2, 1)$. Using the
|
||||
previously described VEM algorithm we obtain for each network its parameters
|
||||
($\rho,\pi,\alpha$).
|
||||
|
||||
We then compute the marginal laws for each dimension, for each network. Then
|
||||
we order the network blocks by the probabilities obtained in decreasing order.
|
||||
\begin{itemize}
|
||||
\item For the memberships on the columns:
|
||||
$col~order_m = order\left(\pi_m \times \alpha_m\right)$
|
||||
\item For the memberships on the rows:
|
||||
$row~order_m = order\left(\rho_m \times ~^{t}(\alpha_m)\right)$
|
||||
\end{itemize}
|
||||
|
||||
Using this order we relabel the memberships for the $M$ fitted collection of a
|
||||
single network.
|
||||
Then we use the $M$ memberships to fit a collection containing the $M$ networks.
|
||||
\subsection{Greedy exploration to find an estimation of the mode}
|
||||
Using the previously fitted models for $Q = (1,2)$ and $Q = (2,1)$ we choose to
|
||||
perform a greedy exploration to find a first mode.
|
||||
|
||||
Meaning that for a given $Q = (Q_1, Q_2)$ we will compute all the possible
|
||||
memberships for the points $Q = (Q_1 + 1, Q_2)$ and $Q = (Q_1, Q_2 + 1)$, fit
|
||||
the corresponding models and choose the one that maximizes the $BIC-L$ as the
|
||||
next point from which to repeat the procedure. We repeat the procedure until the
|
||||
$BIC-L$ stops increasing $3$ times in a row.
|
||||
|
||||
% \begin{algorithm}
|
||||
% \caption{Greedy exploration of the latent space $Q_1$, $Q_2$}
|
||||
% \label{alg:greedy_explotation}
|
||||
% \textbf{Commencer} initialisation
|
||||
% \begin{itemize}
|
||||
% \item Pour chacun des $M$ réseaux, inférer les paramètre avec $Q_1 = q_{1,0}$ et $Q_2 = q_{2,0}$
|
||||
% \item Apparier les clusterings obtenus en utilisant les probabilités marginales, afin de faire correspondre les étiquettes des clusters obtenus.
|
||||
% \end{itemize}
|
||||
% \textbf{tant que} le BICL du meilleur voisin sélectionné à chaque itération n'a pas augmenté durant 3 itérations consécutives, continuer :
|
||||
% \begin{itemize}
|
||||
% \item Calculer toutes les séparations possible de chacun des
|
||||
% \end{itemize}
|
||||
% \end{algorithm}
|
||||
|
||||
When this first estimation of the $BIC-L$ mode has been find we apply the moving
|
||||
window on it.
|
||||
\subsection{Fenêtre glissante pour mettre à jour les clusterings et les $BIC-L$}
|
||||
|
||||
\section{Clustering des réseaux}
|
||||
\subsection{Adaptation de la distance entre les paramètres du modèle}
|
||||
La distance pondère désormais avec les $\pi$ et les $\rho$.
|
||||
\[
|
||||
D_{\mathcal{M}}(m,m') = \sum_{q = 1}^{Q_1} \sum_{r = 1}^{Q_2} \max(\widetilde{\pi}_{q}^{m}, \widetilde{\pi}_{q}^{m'}) \left( \frac{\widetilde{\alpha}_{qr}^{m}}{\widehat{\delta}_{m}} - \frac{\widetilde{\alpha}_{qr}^{m'}}{\widehat{\delta}_{m'}}\right)^{2} \max(\widetilde{\rho}_{r}^{m}, \widetilde{\rho}_{r}^{m'})
|
||||
\]
|
||||
|
||||
\listoffigures
|
||||
\listoftables
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue