Artificial Intelligence - P2

School Project for Artificial Intelligence
This project was divided in 2 parts

Part 1

Tested methods of modelation and inference with Bayesian Networks.

Implemented in python. (file BN.py) toposort.py was given.

Part 2

Implemented Q-learning algorithm, that given a trajectory computes Q-values for each action.

Implemented in python. (file RL.py)
RL.py had some base code.

Report

The report is in portuguese
The major part of the whole project was the report:

Part 1:

• Critical discution of the results found;
• Advanatages and disavantages of the implemented methods;
• Complexity of the solution and possible alternatives.

Part 2:

• Graphic representation of the ambient in which the agent is moving;
• Reward function, optimal policy, impact of each action in each state;
• Complexity, advantages/disavantages and alternatives.

IA_P2.pdf

Tests

mainRL.py tests RL.py. Q1.npz and traj.npz are trajectories used by this test.
mainBN.py tests BL.py

Authors

Joao Coelho
Francisco Santos

Grade

19.5/20