Dive into a comprehensive video explanation of the groundbreaking AlphaGo paper, which details the first AI system to defeat a professional Go player. Explore the intricate components of AlphaGo, including supervised learning policies, reinforcement learning networks, and value networks. Gain a deep understanding of Monte Carlo Tree Search (MCTS) and its application in AlphaGo. Learn about the evaluation process, older techniques, and engineering aspects behind this revolutionary AI system. Discover how neural networks and symmetries play a crucial role in AlphaGo's success, and grasp the context of why conquering the game of Go was considered a significant milestone in artificial intelligence.
AlphaGo - Mastering the Game of Go with Deep Neural Networks and Tree Search - RL Paper Explained
Aleksa Gordić - The AI Epiphany via YouTube
Overview
Syllabus
Intro
Context behind the game of Go
High-level overview of components - SL policies
RL policy network
The value network
Going deeper
Details around value network
Understanding the search MTCS
Evaluation of AlphaGo
Older techniques
Even more detailed explanation of APV-MTCS
Virtual loss
Engineering
Neural networks and symmetries
Taught by
Aleksa Gordić - The AI Epiphany
Related Courses
-
DeepMind's AlphaGo Zero and AlphaZero - RL Paper Explained
-
Write Your Own AI Agent with Monte Carlo Tree Search - Building a Tic-Tac-Toe Game
-
From Tic Tac Toe to AlphaGo - Playing Games with AI and Machine Learning
-
The Role of Multi-Agent Learning in Artificial Intelligence Research at DeepMind
-
Divide-and-Conquer Monte Carlo Tree Search for Goal-Directed Planning - Paper Explained
-
MuZero - Mastering Atari, Go, Chess, and Shogi by Planning with a Learned Model - RL Paper Explained
