Xu, Mme Catherine (2020) Food Recipe Generation with Computational Creativity PFE - Project Graduation, ENSTA.
Gastronomy is an essential part of each country. Through recipes can we better understand their culture. Food recipe generation is therefore a task that requires to first better comprehend the recipes, and to integrate cultural knowledge. We propose an Encoder- Decoder architecture that takes as input the list of ingredients and cuisine and output the recipe, with eventual additional ingredients that would contribute to a better recipe. We developed a Pairing Attention that selects the additional ingredients based on the ingredients in the input and the good pairings with them. We did an extensive cuisine classification work to then experiment with the cuisine as input to the recipe generation model. Our results show that using the cuisine is indeed an improvement over no input or the recipe title as input. However, the Pairing Attention still requires some modifications to give better results than the baseline.
|Item Type:||Thesis (PFE - Project Graduation)|
|Uncontrolled Keywords:||Traitement de language naturel; cuisine informatique; génération de recette; méchanisme d’attention; accord d’ingrédients; classification de cuisine|
|Subjects:||Information and Communication Sciences and Technologies|
|Deposited By:||Catherine Xu|
|Deposited On:||11 déc. 2020 15:49|
|Dernière modification:||11 déc. 2020 15:49|
Repository Staff Only: item control page