Applied Deep Learning 2025 - Lecture 6 - Deep Reinforcement Learning
About this lesson
In this lecture, we're looking at Deep Reinforcement Learning and how machines can learn from experiences and navigate in an environment where you only have limited knowledge. Topics include rewards, the value function, policies, and how deep learning can be used to learn a policy function. Complete Playlist: https://www.youtube.com/watch?v=vlTnIjhhmzA&list=PLNsFwZQ_pkE8H1o874cZbiwnNRJ6hCDJI 00:00:00 - Start 00:00:48 - What is reinforcement learning 00:02:17 - Classes of Learning Problems 00:04:47 - Challenges in Reinforcement Learning 00:09:53 - Key Concepts 00:15:15 - Quality, Value, and Policy Functions 00:18:56 - Q-Function to the Policy 00:20:19 - Deep Reinforcement Learning 00:30:24 - Policy Gradient 00:34:22 - Exploration vs. Exploitation 00:36:03 - Reward Shaping, Auxiliary Tasks, and Hindsight Experience Replay 00:50:03 - Practical Reinforcement Learning 00:52:48 - Pong with Reinforcement Learning 00:57:10 - Frameworks and Libraries 00:59:22 - Examples 01:07:42 - Summary == Literature == 1. Amini, Soleimany, MIT 6.S191 - Introduction to Deep Learning. http://introtodeeplearning.com/ 2. rlpyt: A Research Code Base for Deep Reinforcement Learning in PyTorch: https://arxiv.org/abs/1909.01500 3. A Survey of Reinforcement Learning Informed by Natural Language: http://arxiv.org/abs/1906.03926v1 4. Stanford CS231n: Reinforcement Learning in Visual Computing https://www.youtube.com/watch?v=lvoHnicueoE 5. Stanford CS234: Reinforcement Learning https://www.youtube.com/playlist?list=PLoROMvodv4rOSOPzutgyCTapiGlY2Nd8u 6. Deep Mind: Reinforcement Learning with Unsupervised Auxiliary Tasks, 2016 https://arxiv.org/abs/1611.05397 7. Arxiv Insights: Overcoming spare rewards in Deep RL, 2018. https://youtu.be/0Ey02HT_1Ho 8. Pathak et al. Curiosity-driven Exploration by Self-supervised predicition, 2017. https://arxiv.org/abs/1705.05363 9. Andrychowicz et al. Hindsight Experience Replay, 2017. https://arxiv.org/abs/1707.01495 10. AlphaStar: Mastering the Real-Time Strat
DeepCamp AI