# Anfan

> Reinforcement learning for interactive video game AI.

Anfan studies how game agents learn from observations, actions, and feedback, with a focus on long-horizon learning, generalization, adaptation, and rigorous evaluation.

## Selected projects

- [Combat Learning Agent](https://ericxucui.github.io/CombatLearningAgent/): A UE5 reinforcement-learning agent trained with PPO for real-time melee combat.
- [Decision Scheduling for RL Game Agents](https://ericxucui.github.io/#current-work--decision-scheduling-for-rl-game-agents): Research on when real-time RL agents should make their next decision as actions and interruptions unfold.
- [Motion-Controlled Medieval Combat](https://github.com/EricXuCui/UE4-CPP-Motion-Controlled-Melee-Combat-System): An Unreal Engine C++ prototype using gyroscope input for multiplayer melee combat.

## Research focus

- Reinforcement learning for sequential decision-making in video games.
- Generalization across changing game conditions.
- Adaptation to new situations during play.
- Reproducible evaluation and failure analysis.

## Site

- [Anfan home](https://anfan.org/): Research agenda and selected projects.
- [Anfan home in Markdown](https://anfan.org/index.md): Generated Markdown representation of the home page.

## Contact

- [Research email](mailto:research@anfan.org)
- [General email](mailto:hello@anfan.org)