Machine learing is an umbrella term for certain types of learning
algorithms in artificial intelligence. The purpose of machine learning
is to make programs learn to performe tasks in accordance to certain
given conditions. Reinforced learning is an area within machine learning
which is all about making the program learn what is a good action by
being rewarded.
One algorithm based on reinforcement learning is the Q-learning
algorithm. It uses a table containing values for each state it has
encountered. The table is updated with new values when new rewards are
awarded by actions it has performed in the environment. This report
describes how Q-learning can be implemented an how our implemention
behaves against three different adversaries.
The purpose of this report is to investigate how the Q-learning learning
vaiable α affects the rate of learning and for which values this
implementation can performe optimaly. Our best result was α=0.9.