Markov Decision Process for Industrial Optimization and Overview of TIBCO Data Science platform

Reinforcement learning enables systems to perform decision making upon the agent-environment interaction. It has gained pace in adoption and Markov Decision Process are builting blocks of RL based optimization. In the presentation , we will discuss about the MDP based solution that operate on Policy and Value Iteration and Q Learning. MDPtoolbox is a package offered in R to implement RL solutions.There are multiple optimization methods being used in the industry to address optimization problems but as the size of the problem grows most of the methods become inefficient in performance and practicality.MDP solutions are good at sub-problem solutions as well and so can be used in an online mode.

Later , we will discuss about the TIBCO Data Science platform and how its enabling business world wide to operationalize data science application . Its a suite of TIBCO Spotfire , TIBCO Statistica , TIBCO Alpine, TIBCO Streambase and TERR. The Gartner report for 2019 has listed TIBCO Data Science as leader in ML platforms.


Outline/Structure of the Talk

- Introduction to Markov Decision Process
- Methods to solve MDP
- Use case of MDP based solution
- How TIBCO Data Science platform helps enterprise to implement and operationalize data science.

Learning Outcome

- How to use MDP to solve industrial optimization

- Solving Stochastic Optimization using Policy Iteration , Value Iteration and Q Learning

- TIBCO Data Science platform overview and capability

Target Audience

ML engineers , Business Analyst & Program Managers

Prerequisites for Attendees

There is no prerequisite . Beginner level familiarity with ML and optimization is a plus.

schedule Submitted 1 year ago

Public Feedback