Reinforcement Learning and Minecraft - ML Con Spring 2018

Overview

Explore reinforcement learning through Minecraft's AI platform, Project Malmo. Learn to solve problems using deep learning techniques in a unique experimental environment.

Syllabus

• Introduction to Reinforcement Learning
Overview of Reinforcement Learning (RL)
Key Concepts: States, Actions, Rewards
Exploration vs. Exploitation
• Project Malmo and Minecraft as a Platform for AI
Introduction to Project Malmo
Setting Up the Environment
Understanding the Minecraft Game Environment
• Fundamentals of Deep Learning
Neural Networks Basics
Deep Q-Networks (DQN)
Policy Gradient Methods
• Implementing Reinforcement Learning in Project Malmo
Building and Training Agents
Experimentation with Different Strategies
Case Study: Solving a Task in Minecraft
• Advanced Techniques and Applications
Exploration Strategies in RL
Transfer Learning in RL
Multi-Agent Reinforcement Learning
• Evaluating and Improving RL Models
Performance Metrics for RL
Hyperparameter Tuning
Troubleshooting Common Problems
• Project and Practical Application
Define a Problem to Solve in Minecraft
Design and Implement a Solution
Presenting Results and Findings
• Future Directions in AI and Reinforcement Learning
Recent Advancements
Ethical Considerations in RL
Future Applications in Gaming and Beyond

---

Subjects

Conference Talks