2021/06/15 by Péter Pfeiffer, Pfeiffer, Peter, Johannes Lahann +3 · 1 citation
Business, Management and Accounting · Decision Sciences · #Artificial Intelligence (cs.AI) #Big Data and Business Intelligence #Business Process Modeling and Analysis #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Stock Market Forecasting Methods
paper · pdf · doi:10.48550/arxiv.2106.08027
openalex publication_date 2021/06/15 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Learning meaningful representations of data is an important aspect of machine\nlearning and has recently been successfully applied to many domains like\nlanguage understanding or computer vision. Instead of training a model for one\nspecific task, representation learning is about training a model to capture all\nuseful information in the underlying data and make it accessible for a\npredictor. For predictive process analytics, it is essential to have all\nexplanatory characteristics of a process instance available when making\npredictions about the future, as well as for clustering and anomaly detection.\nDue to the large variety of perspectives and types within business process\ndata, generating a good representation is a challenging task. In this paper, we\npropose a novel approach for representation learning of business process\ninstances which can process and combine most perspectives in an event log. In\nconjunction with a self-supervised pre-training method, we show the\ncapabilities of the approach through a visualization of the representation\nspace and case retrieval. Furthermore, the pre-trained model is fine-tuned to\nmultiple process prediction tasks and demonstrates its effectiveness in\ncomparison with existing approaches.\n