Implementations of various ML algorithms for playing the game of bomberman.
The repository contains (in addition to the original repository) the following files:
.
├── agent_code
│ : # binary agents (features are binary vectors)
│ ├── binary_agent
│ ├── binary_agent_v1
│ ├── binary_agent_v2
│ ├── binary_agent_v3
│ ├── binary_agent_v4
│ ├── binary_agent_v5
│ ├── binary_agent_v6
│ : # binary distance agents
│ : # same as binary but values are 1/distance
│ ├── binary_distance_agent_v1
│ ├── binary_distance_agent_v2
│ : # distance agent (features are distances)
│ : # differs from binary_distance (gives all distances, not just shortest)
│ ├── distance_agent
│ ├── distance_agent_v1
│ : # q-table agents
│ ├── q_agent
│ ├── q_agent_v1
│ ├── q_agent_v2
│ └── q_agent_v3
├── graphs # matplotlib graphs for the agent report
├── elo # elo data
├── elo.py # script for calculating elo
├── train.py # script for training agents
└── train_q.py # old script for training q-table