AI Driven Automation for "Classical" MDP

We consider the problem of application of "classical" Markov Decision Process (MDP) to medium-size problems such as waste-water treatment plants and multi-echelon supply chain. These problems consists of dozens to hundreds of state variables and suffer from the "curse of dimensionality". Furthermore, these variables, especially continuous ones, require binning which is time consuming. We show how AI provides automation that lets the LP based MDP algorithm to be useful for solving real-life problems.