What you would learn in Advanced Reinforcement Learning: policy gradient methods course?
This is the complete Reinforcement learning course series available on Udemy. In this course, you'll be taught how to implement the most efficient Deep Reinforcement Learning algorithms in Python using PyTorch and PyTorch lightning. Then, you will create the creation of adaptive algorithms to tackle control tasks by relying on previous experiences. Learn to combine these methods using Neural networks and Deep Learning methods to create adaptable Artificial Intelligence agents capable of taking on decision-making tasks.
In this course, you will be introduced to the latest developments in Reinforcement learning techniques. This course will prepare students for subsequent courses in the series, which will look at other techniques that are effective in different kinds of tasks.
The class is focused on improving practical capabilities. Thus, after learning the fundamental aspects of each family of methods, we'll apply any of them within Jupyter notebooks beginning from scratch.
Modules for leveling:
Refresher: The Markov decision-making procedure (MDP).
- Refresher: Q-Learning.
Refresher: Short Introduction to Neural Networks.
Advance Reinforcement Learning:
- PyTorch Lightning.
- Hyperparameter tuning via Optuna.
Reinforcement Learning using images as inputs
Double Deep Q-Learning
Duplicating Deep Q Networks
Prioritized Experience Replay (PER)
" Distributional Deep Q-Networks
- Noisy Deep Q-Networks
N-step Deep Q-Learning
Rainbow Deep Q-Learning
Learn the most sophisticated Reinforcement Learning techniques.
Learn to build AIs that operate in a complex setting to accomplish their objectives.
Develop from scratch the most advanced Reinforcement Learning agents with Python's most widely used instruments (PyTorch Lightning OpenAI gymand Optuna)
Learn how to do hyperparameter tuning (Choosing the most appropriate experimental environment for AIs to study)
Understand the learning process of each algorithm.
Expand and debug the algorithms described.
Learn and apply new algorithms that are derived from research papers.