Get updates on Eligibility, Admission, Placements Fees Structure
Quick Facts
| Medium Of Instructions | Mode Of Learning | Mode Of Delivery |
|---|---|---|
| English | Self Study | Video and Text Based |
Course Overview
The Practical Reinforcement Learning programme by Coursera is the fourth out of the seven courses included in the ‘Advanced Machine Learning Specialization’. Offered by the National Research University- Higher School of Economics, this online programme will make you an expert in the field of machine learning.
Reinforcement Learning is one of the fundamentals of machine learning. Without getting a good grasp of this particular area, one cannot master the art of artificial intelligence. With the certification course in Practical Reinforcement Learning by Coursera, you can get in-depth knowledge about Reinforcement Learning and hone the various skills required to master this area of machine learning.
The Practical Reinforcement Learning online course by Coursera is well equipped with a planned curriculum as well as offerings like graded quizzes and assignments, peer feedbacks, practice quizzes, etc. that will help you to become proficient in this field. You will earn a shareable certificate upon completion as well.
The Practical Reinforcement Learning online programme by Coursera is a full package that covers all the core concepts of Reinforcement Learning, including free methods, dynamic programming, value-based methods, neural networks, algorithm-writing, model-free methods, and more.
The Highlights
- Full course specialization
- Course videos and readings
- 100% online
- Self-paced learning
- Shareable certificate
- Completion in approximately 26 hours
- Offered by National Research University - Higher School of Economics
Programme Offerings
- Online videos and readings
- Graded Quizzes
- Grades assignments
- Programming Assignments
- Full Specialization
- Self-paced learning
- peer feedback
Courses and Certificate Fees
| Certificate Availability | Certificate Providing Authority |
|---|---|
| yes | Coursera |
Interested applicants can opt for a 7-day free trial and enrol for free. To avail full contents of the course along with the certificate, candidates need to purchase the experience certificate.
What you will learn
Once you complete the Practical Reinforcement Learning program, you will:
- Build a strong foundation of reinforcement learning methods like value/policy iteration, policy, gradient, q-learning, etc.
- Understand the functioning and usage of deep neural networks for tasks in reinforcement learning, also called the “hype train”
- Acquire skills to write and develop state-of-the-art algorithms that make the machines behave in the desired way
- Know how to teach neural networks to play games through reinforcement learning, and also use it for contextual bandits and seq2seq
Who it is for
Admission Details
Follow the steps given below to take admission in the Practical Reinforcement Learning program by Coursera :
- Visit the Course page.
- On the website, glide your cursor over the “Explore” button. Under the “Data Science” category, choose “Machine Learning”.
- This will display all the courses offered in Machine Learning. Choose the Course you want to take admission. In this case, choose “Practical Reinforcement Learning”.
- You will be redirected to the course-specific webpage. Locate the “Enroll for Free” button and click on it.
- Sign up using your Email ID and password. If you already have an account, use the details to log in. You will get a 7-day free trial once you enrol.
- To get the full course offerings and continue after the free trial, you will have to purchase the certificate experience. You can choose your preferred payment to do so.
Application Details
There is no additional application form for the course. Interested candidates are required to sign up using their Email ID and password, or their Google, Facebook, or Apple accounts.
The Syllabus
- About the University
- Why should you care
- More on the approximate cross-entropy method
- Reinforcement learning vs all
- Approximate cross-entropy method
- Multi-armed bandit
- Decision process & applications
- Markov Decision Process
- Cross-entropy method
- Evolution strategies: the core idea
- Evolution strategies: duct tape
- Evolution strategies: math problems
- Evolution strategies: log-derivative trick
- Blackbox optimization: drawbacks
- Measuring Policy Optimality
- Policy: evaluation & improvement
- Policy and value iteration
- State and Action Value Functions
- Reward design
- Model-based vs model-free
- On-policy vs off-policy; Experience replay
- Monte-Carlo & Temporal Difference; Q-learning
- Footnote: Monte-Carlo vs Temporal Difference
- Exploration vs Exploitation
- Accounting for exploration. Expected Value SARSA
- Supervised & Reinforcement Learning
- Partial observability
- Double Q-learning
- Loss functions in value-based RL
- Difficulties with Approximate Methods
- DQN – bird's eye view
- DQN – the internals
- DQN: statistical issues
- More DQN tricks
- Intuition
- All Kinds of Policies
- Policy gradient formalism
- The log-derivative trick
- REINFORCE
- Advantage actor-critic
- Duct tape zone
- Policy-based vs Value-based
- Case study: A3C
- Combining supervised & reinforcement learning
- Recap: bandits
- Regret: measuring the quality of exploration
- The message just repeats. 'Regret, Regret, Regret’.
- Intuitive explanation
- Thompson Sampling
- Optimism in the face of uncertainty
- UCB-1
- Bayesian UCB
- Introduction to planning
- Monte Carlo Tree Search