November 3
@
8:00 am
–
December 15
@
5:00 pm
Reinforcement Learning
In this course, “Reinforcement Learning”, you will learn about various reinforcement learning (RL) algorithms, a branch of machine learning and AI. This course covers everything you need to know about RL, including an overview of the basic concepts of RL, value-based methods, policy-based methods, and actor-critic algorithms. You will also learn how to leverage these algorithms to address specific real-world problems.
In this course, you will engage in hands-on activities, homework, and instructor consulting to make learning RL enjoyable and rewarding. You will also be able to tackle real-world problems in discrete and continuous controls, and sequential decision-making. By the end of this course, you’ll have the skills and confidence to tackle any machine-learning challenge with RL.
Course At a Glance
Instructional Period: November 3 – December 15, 2025
Time to Complete Badge: 60 days
Last Dy to Earn Badge: December 15, 2025
Expertise Level: Intermediate/Advanced
Scientific Machine Learning
Parallelism in Deep Learning
- Basic Python programming
- Basic understanding of deep learning
- Basic understanding of control and sequential decision-making
- Basic PyTorch programming
Audience: The course is intended for a broad audience within the spectrum of the software and technology industry, including software engineers, data scientists, data engineers, data analysts, research scientists, and software developers. We have had past students from Corteva, Bayer, John Deere, Principal, Jord Biotech, Ag Leader, and other leading organizations. The course is designed to provide comprehensive understanding of reinforcement learning and how to use these methods for a broad range of audiences and applications.
By the end of the course, you should be able to:
Develop policy gradient algorithms with deep neural networks implementing advanced policy gradient methods and training both value and policy networks for complex control tasks
Formulate reinforcement learning problems into Markov Decision Processes by defining state and action spaces, reward functions, and transition dynamics for specific tasks
Develop basic value-based and policy-based reinforcement learning algorithms using deep learning packages such as PyTorch to solve decision-making tasks
Develop Q-learning algorithms with deep neural networks including custom Double Deep Q Networks for discrete and continuous control problems such as CarPole, Pendulum, and Atari Games
- 2 Quizzes covering basic and advanced definitions and concepts of reinforcement learning, including Markov Decision Processes and advanced RL methods
- 2 Coding Exercise Assignments in which you will implement Python codes based on hands-on activities, including: Coding value-based & policy-based RL algorithms to solve classic control problems
Module 1: Introduction to Markov Decision Process and Reinforcement Learning
- Fundamentals: state, action, transition dynamics, reward function, and MDP formulation
- Overview of reinforcement learning framework and problem formulation techniques
- Implementation of GridWorld problem using tabular RL methods
Module 2: Develop Value-based Reinforcement Learning Algorithms
- State value and state-action value functions and their approximations
- Traditional value-based methods: Q-learning and SARSA algorithms
- Deep Q-learning and advanced value-based deep RL algorithms for continuous control
Module 3: Develop Policy-based Reinforcement Learning Algorithms
- Policy gradient methods and deep policy optimization algorithms
- Vanilla policy gradient and Proximal Policy Optimization (PPO) implementation
- Application to continuous control tasks using PyTorch
Module 4: Advanced Deep Reinforcement Learning Methods
- Model-based reinforcement learning methods and applications
- Offline reinforcement learning techniques and their practical applications
- Advanced optimization and best practices for deep RL systems
The course starts on November 3, 2025. All coursework must be completed by December 15, 2025 (possible extension till December 31, 2025 for industry professionals), in order to earn the micro-credential badge. You will continue to have access to the course materials until January 31, 2026. The approximate time to complete this course is 16 hours.
This course has an instructional period from November 3 to December 7, 2025. During this instructional period, course materials will be released weekly and live synchronous sessions will be held. You may complete the course materials at your own pace. Live Zoom meetings will be conducted for interactive coding sessions. A suitable time for these live sessions will be determined through a group poll. The recordings of those sessions will be available soon after each meeting.
You will receive the micro-credential badge upon successful completion of the course assessments.
Course materials are provided within the course. No additional purchase is required.
Registration
Register for the course as a 1-credit independent study course with a maximum of 3 such TrAC courses per semester.
Students must register for the independent study by emailing benearl@iastate.edu and cc-ing baditya@iastate.edu. Also, fill this Google form for our records: https://forms.gle/mQRefUJ4qpa29s5w6
Industry Professionals/ISU Staff/Post Docs
$
500
.00
ISU Professionals/Staff and Government Employees: $300
Zhanhong Jiang, Data Scientist
Zhanhong Jiang is a data scientist in the Translational AI Center (TrAC) at Iowa State University. His research interests lie in decentralized deep learning, reinforcement learning, time-series prediction, and applications to cyber-physical systems. Prior to that, he was a senior AI scientist at Johnson Controls and worked on smart and healthy building solutions using AI/ML technologies. He has numerous publications in prestigious journals and conferences and more than 10 patents.
