SML: Semantic Machine Learning Model Ontology - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

SML: Semantic Machine Learning Model Ontology

Résumé

Artificial Intelligence is a set of technologies that simulate human-like cognition, using computer software and systems, to perform tasks associated with intelligent beings. One method of doing so, is Machine Learning (ML), which enhances system efficiency based on learning algorithms that create models from data and its underlying patterns. Nowadays, many ML models are being generated with different characteristics (e.g., type of the algorithm used, data set used to train it, resulting model performance), thus making the selection of a suitable model for a given use case a complex task, especially for non-expert users (with no or limited knowledge in ML). In this paper, we propose SML, an ontology-based model for Semantic Machine Learning description. SML allows, mainly, to describe and store ML models' characteristics with their operational specifications, related data features, contextual usage, and evaluation metrics/scores to facilitate and improve ML model selection. The conducted experiments show promising results on both the efficiency and the performance levels.
Fichier principal
Vignette du fichier
SML-2023.pdf (1.1 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04270203 , version 1 (03-11-2023)

Identifiants

Citer

Lara Kallab, Elio Mansour, Richard Chbeir. SML: Semantic Machine Learning Model Ontology. International Conference on Web Information Systems Engineering, Oct 2023, Melbourne (AUS), Australia. pp.896-911, ⟨10.1007/978-981-99-7254-8_70⟩. ⟨hal-04270203⟩

Collections

UNIV-PAU LIUPPA
20 Consultations
78 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More