An Introduction to Reinforcement Learning - Snake Your Way Out of a Paper Bag

Learn reinforcement learning fundamentals through hands-on C++ implementation in this 56-minute conference talk from ACCU 2025. Explore how reinforcement learning differs from other AI approaches like LLMs by building an agent that learns to play the classic Snake game through trial and error, with better outcomes rewarded and worse outcomes penalized. Discover the core concepts of machine learning and reinforcement learning while implementing vanilla Q-learning using a simple data table approach, starting with std::vector and progressing through increasingly sophisticated C++ techniques. Understand how reinforcement learning applies to real-world scenarios including targeted marketing, infrastructure optimization, and financial decision-making, while gaining practical experience with probabilistic decision-making and random number generation. Master the fundamentals that can later extend to deep Q-learning with neural networks, and appreciate how implementing algorithms yourself provides the best path to understanding both the programming language and the underlying concepts being implemented.