Pattern recognition in process models using convolutional neural networks

M. Stavtsev, S. Shershakov

HSE University, Moscow, Russia

Abstract: Modeling is utilized to analyze complex processes, identify errors within them, and facilitate their improvement. Among various process modeling approaches, process mining techniques are particularly noteworthy. Process mining algorithms reconstruct process models from event logs by identifying regularities in the form of patterns. Approaches to synthesis include purely algorithmic methods as well as those based on statistical regularities. This research investigates statistical approaches for extracting process model patterns using convolutional neural networks, a defining feature of these networks is their ability to perform sequential feature extraction on the dataset under investigation.

Key words and phrases: patterns, process mining, Image data engineering, CNN.

UDC: 004.852

Received: 28.02.2025

Language: English