-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Fitzwilliam Darcy
authored and
Fitzwilliam Darcy
committed
Feb 1, 2024
1 parent
70ced5b
commit f29e7d6
Showing
10 changed files
with
103 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,49 @@ | ||
--- | ||
title: Phénomène de sur-apprentissage | ||
author: Michel Jean Joseph Donnet | ||
--- | ||
|
||
|
||
# Sur-apprentissage | ||
|
||
On ne veut pas apprendre le bruit des données d'apprentissage ! | ||
|
||
# Sur-apprentissage: exemple | ||
|
||
![larevueia.fr](../res/overfitting_1.png){width=70%} | ||
|
||
# Sur-apprentissage: graphique | ||
|
||
![larevueia.fr](../res/overfitting_2.png){width=70%} | ||
|
||
# Comment éviter le sur-apprentissage ? | ||
|
||
Validation croisée ! | ||
|
||
![datascientest.com](../res/crossvalidation.png) | ||
|
||
# Autres techniques ? | ||
|
||
- Ajout données d'apprentissage modifiées (pour plus de généralisation...) | ||
- Retirer des caractéristiques | ||
- ... | ||
|
||
# Concrêtement, dans le projet | ||
|
||
Dans le projet, pour montrer le phénomène de sur-apprentissage: | ||
|
||
- Ajout de bruits aux données d'apprentissage | ||
- Volume réduit de données | ||
- Modification du nombre d'itérations | ||
|
||
# Résultats obtenus | ||
|
||
\center ![](../res/overfitting.png){width=80%} | ||
|
||
# Résultats obtenus | ||
|
||
\center ![](../res/overfitting_reg.png){width=80%} | ||
|
||
# Résultats obtenus | ||
|
||
\center ![](../res/overfitting_reg_2.png){width=80%} |
Binary file renamed
BIN
+1.06 MB
presentation/Michel_Donnet_overfitting.pdf → presentation/Overfitting_MJJD.pdf
Binary file not shown.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,54 @@ | ||
--- | ||
title: Régression logistique | ||
author: Michel Donnet | ||
--- | ||
|
||
|
||
# Régression logistique binaire: principe | ||
|
||
![](../res/decision.png) | ||
|
||
# Régression logistique binaire: idée | ||
|
||
\center ![](../res/pomme.png){width=80%} | ||
|
||
# Régression logistique: fonction d'estimation | ||
|
||
Fonction sigmoïde | ||
$$\sigma(z) = \frac{1}{1 + e^{-z}}$$ | ||
|
||
Caractérisiques: | ||
|
||
- Comprise entre 0 et 1 $\Rightarrow$ probabilité ! | ||
- Point d'inflexion à 0.5 | ||
|
||
Idée: | ||
|
||
- établir un seuil afin de prédire le label $Y$ | ||
|
||
# Entraînement du modèle | ||
|
||
But: | ||
|
||
- maximiser la probabilité $P(Y = y | X)$ pour $y$ la valeur d'entrainement du label. | ||
|
||
Mais on a la descente en gradient... | ||
|
||
$\Rightarrow$ transformer le problème en problème de minimisation ! | ||
|
||
$\Rightarrow$ `Negative Logarithm Likelihood` | ||
|
||
# Régression logistique multinomiale: principe | ||
|
||
![](../res/directions.png) | ||
|
||
# Généralisation de la fonction sigmoïde en fonction softmax | ||
|
||
$$P(Y = k | X) = \frac{1}{1 + e{-X \theta^T}}\ \rightarrow \ \frac{e^{X \theta_k^T}}{\sum_i^N e^{X \theta_i^T}}$$ | ||
|
||
# Entraînement du modèle | ||
|
||
Même principe que pour la régression logistique binaire | ||
|
||
|
||
|
Binary file renamed
BIN
+568 KB
presentation/Michel_Donnet_Regression.pdf → presentation/Regression_MJJD.pdf
Binary file not shown.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.