svpino/cs7641-assignment4

CS7641 - Machine Learning - Assignment 4 - Markov Decision Processes

cs7641-assignment4's Issues

Store agent state and reward value

How can I store the agent state steps and the reward value among program executions? I mean in each episodes. I want to use it for the analysis. I found in Burlap explanation that we can use writeQTable and loadQTable to store it, however I don't understand how to do it in your code. I am sorry I am very new to Burlap and Java. Thank you.

Goal Reward is ignored

No matter what I set the goal reward to, it remains at 0 for intents and purposes. I even tried making the goal reward negative, as can be seen in the image, but the policy still targets the goal, no matter what.