Deep reinforcement learning for predictive monitoring under LTLf constraints

Efrén Rama Maneiro

PhD student at the University of Santiago de Compostela

Predictive monitoring is a subfield of process mining that focuses on predicting how a process will unfold. Deep learning techniques have become popular in this field due to their enhanced performance with respect to classic machine learning models. However, most of these approaches for predictive monitoring overlook the treatment of time, understanding “time” as the relationships between the activities of the process. To improve the performance of deep business predictive models, this research proposal aims to define a novel predictive monitoring approach that learns from a mined set of restrictions expressed in temporal logic using deep reinforcement learning. By improving the treatment of time in deep learning models for predictive monitoring, this research has the potential to significantly open new research avenues in the field of business process management.

Visit to: Dipartimento di Ingegneria Informatica, Automatica e Gestionale. Sapienza Università di Roma. Host scientists: Fabio Patrizi, Andrea Marrella

More information about Efrén: https://scholar.google.es/citations?user=JshQ_5AAAAAJ&hl=es