Learning optimal policies from event logs through reinforcement learning: a comparison of deep and MDP-based approaches
A new research paper explores learning optimal policies for prescriptive process monitoring using reinforcement learning. The study compares a model-based approach using Markov Decision Processes (MDPs) with a model-free Deep Reinforcement Learning (DRL) method. Both techniques aim to learn intervention strategies directly from historical event data, minimizing the need for domain knowledge and handling scenarios with external actors. While both methods showed similar effectiveness in improving Key Performance Indicators (KPIs), the MDP-based approach was found to be more computationally efficient. AI
IMPACT This research could lead to more data-driven and automated optimization of business processes by learning intervention strategies directly from historical data.