# Schedule
This is a directed study in Computer Science. In Spring 2024, we will work on a project on training AI agents (NPCs) in video games, which involves reinforcement learning (to explore the environment and learn optimal actions), game theory (to compute optimal responses to other NPCs), and some computer graphics (to simulate the environment).
We ended up solving a simple grid world hide-and-seek problem, a very simple version of
Link, and we observed interesting and un-intuitive agent behavior that exploits specific implementations of the environment.
| Week |
Date |
Topic |
Notes |
| 1 |
Jan 24 |
Reinforcement Learning |
W1 |
| 2 |
Jan 31 |
Markov Decision Process |
W2 |
| 3 |
Feb 7 |
Q Learning |
W3 |
| 4 |
Feb 14 |
Game Theory |
W4 |
| 5 |
Feb 21 |
Markov Game |
W5 |
| 6 |
Feb 28 |
Markov Perfect Equilibrium |
W6 |
| 7 |
Mar 6 |
Neural Networks |
W7 |
| 8 |
Mar 13 |
Genetic Algorithm |
W8 |
| 9 |
Mar 20 |
Deep Q Network |
W9 |
| 10 |
Mar 27 |
- |
W10 |
| 11 |
Apr 3 |
Computer Graphics |
W11 |
| 12 |
Apr 10 |
3D Transformations |
W12 |
| 13 |
Apr 17 |
Collision Detection |
W13 |
| 14 |
Apr 24 |
Environment Simulation |
W14 |
| 15 |
May 1 |
- |
W15 |
References:
Link,
Link.
Textbooks: AI for Games:
Link, Reinforcement Learning:
Link, Game Theory:
Link, Multi-Agent Systems:
Link, Computer Graphics:
Link.
Programming Language: JavaScript, TensorFlow.js:
Link, Three.js:
Link.