Sustainability and explainability through learning on large knowledge graphs

When:
05/05/2024 all-day
2024-05-05T02:00:00+02:00
2024-05-05T02:00:00+02:00

Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : Mines Saint-Étienne / LIMOS
Durée : 3 ans
Contact : victor.charpenay@emse.fr
Date limite de publication : 2024-05-05

Contexte :
Large Language Models (LLMs), characterized by a large number of parameters and/or a large number of symbols in their training corpus, have become a reference in the development of AI systems. Yet, their use implies a significant energy consumption , both during training and inference, and a lack of transparency about decisions made by the system. The goal of the thesis will be to show that Knowledge Graphs (KGs), such as DBpedia, BabelNet or ConceptNet can be a solution to both problems.

Sujet :
The thesis will consist in characterizing the relationship between performances of a KG embedding model and its computational cost—empirical laws do exist for LLMs—and in analyzing the correspondance between its geometric properties and the semantic properties of the KG.

Profil du candidat :
Applicants should have prior experience with Semantic Web technologies and/or logic programming. General knowledge about machine learning techniques is also recommended.

Formation et compétences requises :
Master’s degree in computer science, data science or any related topic.

Adresse d’emploi :
Saint-Étienne

Document attaché : 202401231512_lkm-en.pdf