Description
Date depot: 5 avril 2024
Titre: Contrastive Representation Learning for Anomaly Detection on Attributed Networks
Directeur de thèse:
Mohamed NADIF (Centre Borelli (EDITE))
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Intelligence artificielle
Resumé: The primary objective of the thesis project is to explore a cutting-edge joint methodology that merges contrastive representation learning with anomaly detection within attributed networks. The project also seeks to develop a novel and flexible mathematical model that enables the simultaneous representation of data and detection of anomalies. This model will be designed to incorporate additional data seamlessly, such as connections between nodes or constraints on data points into a cohesive learning framework. The ultimate goal is to enhance the model's ability to identify multiple anomalies at once, providing effective solutions to current challenges in detection capabilities. Moreover, this project encompasses a research methodology that spans both theoretical foundations and practical applications, targeting areas such as cybersecurity, natural language processing (NLP), and multi-omics data.