Agent57 - Outperforming the Atari Human Benchmark
Offered By: Yannic Kilcher via YouTube
Course Description
Overview
Explore DeepMind's groundbreaking Agent57, the first reinforcement learning agent to surpass human performance across all 57 Atari benchmark games. Delve into the innovative extensions of previous algorithms like Never Give Up and R2D2, focusing on the meta-learning approach for balancing exploration and exploitation. Examine the adaptive mechanism that prioritizes policies throughout training and the novel neural network architecture enabling more stable learning. Gain insights into the significance of this achievement for the field of artificial intelligence and its implications for future advancements in reinforcement learning.
Syllabus
Agent57: Outperforming the Atari Human Benchmark
Taught by
Yannic Kilcher
Related Courses
Creative Acts of Curious PeopleCreativeLive Hiper-Memória & Hiper-Aprendizagem
Udemy Steps to AGI
Edan Meyer via YouTube Acquiring and Understanding Cross-Task Generalization with Diverse NLP Tasks
USC Information Sciences Institute via YouTube AutoML + ML APIs = Machine Learning for All Developers
Devoxx via YouTube