Deep Reinforcement Learning

This course provides a complete introduction to deep reinforcement learning, where we connect reinforcement learning methods with deep neural networks. The focus is both on understanding the concepts and on building working implementations step by step.We start by reviewing the basics of reinforcement learning and how function approximation works with neural networks. From there, we move into value-based methods like Deep Q-Networks (DQN) and their extensions, policy gradient algorithms such as PPO, DDPG, TD3, and SAC, and advanced techniques for exploration, model-based learning, and multi-agent training.The course is hands-on, with practical coding exercises in PyTorch. You will build your own agents, experiment with environments like Atari and robotics simulations, and learn how to set up a proper development pipeline for deep reinforcement learning research and applications.In addition to the core algorithms, we cover important modern topics including curiosity-driven exploration, attention mechanisms, world models, distributed training, and reinforcement learning from human feedback. These topics will give you a broader perspective on how deep reinforcement learning is applied in practice today.By the end of the course, you will be able to:Understand how reinforcement learning and deep learning combineImplement key algorithms from scratch and run experimentsApply advanced methods to complex environmentsUse tools like PyTorch and Gymnasium to develop and test your own ideasThis course is designed for learners who already know the basics of reinforcement learning and are comfortable with linear algebra and Python. If you want to expand your skills into modern DRL methods, this course will give you the foundation and practice you need.