90 lines
1.7 KiB
Markdown
90 lines
1.7 KiB
Markdown
|
---
|
||
|
|
theme: academic
|
||
|
|
class: text-white
|
||
|
|
coverAuthor: Laurent Fainsin
|
||
|
|
coverBackgroundUrl: https://git.fainsin.bzh/ENSEEIHT/projet-fin-etude-rapport/media/branch/master/assets/aube.jpg
|
||
|
|
coverBackgroundSource: Safran Media Library
|
||
|
|
coverBackgroundSourceUrl: https://medialibrary.safran-group.com/Photos/media/179440
|
||
|
|
coverDate: '2023-09-07'
|
||
|
|
themeConfig:
|
||
|
|
paginationX: r
|
||
|
|
paginationY: t
|
||
|
|
paginationPagesDisabled:
|
||
|
|
- 1
|
||
|
|
title: Projet de fin d'étude
|
||
|
|
---
|
||
|
|
|
||
|
|
<h2 class="opacity-50" style="font-size: 2rem;">Projet de Fin d'Étude</h2>
|
||
|
|
|
||
|
|
<h1 style="font-size: 2.4rem; line-height: normal;">Modèles génératifs pour la représentation latente d'aubes 3D sous forme de maillages non structurés</h1>
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Sommaire
|
||
|
|
|
||
|
|
<div class="h-100 flex items-center text-2xl">
|
||
|
|
|
||
|
|
- Présentation de Safran
|
||
|
|
- Modèles génératifs
|
||
|
|
- Présentation du dataset
|
||
|
|
- Génération par diffusion
|
||
|
|
- Résultats
|
||
|
|
- Conclusion
|
||
|
|
|
||
|
|
</div>
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Présentation rapide de Safran
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Modèles génératifs (traditionnels)
|
||
|
|
|
||
|
|
<img src="https://git.fainsin.bzh/ENSEEIHT/projet-fin-etude-rapport/media/branch/master/assets/online_adaptative_sampling_DOE.png" class="m-auto h-100 mt-10"/>
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Modèles génératifs (deep learning)
|
||
|
|
|
||
|
|
<img src="https://lilianweng.github.io/posts/2021-07-11-diffusion-models/generative-overview.png" class="m-auto h-110"/>
|
||
|
|
|
||
|
|
<a href="https://lilianweng.github.io/posts/2021-07-11-diffusion-models/" class="absolute bottom-0 font-extralight mb-1 mr-2 right-0 text-xs">Lilian Weng, 2021</a>
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Dataset
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Denoising Diffusion Probabilistic Models (DDPM)
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Forward process
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Reverse process
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Latent Diffusion Models (LDM)
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Classifier-free Guidance (CFG)
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Résultats
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Vérification par Gaussian Processes (GP)
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Conclusion
|
||
|
|
|