<
Previous

Next

Contents
>
Read this tutorial comfortably offline. Click here to purchase the complete Ebook of this tutorial
QLearning Algorithm
Our virtual agent will learn through experience without teacher (this is called unsupervised learning). The agent will explore state to state until it reaches the goal. We call each exploration as an
episode
. In one episode the agent will move from initial state until the goal state. Once the agent arrives at the goal state, program goes to the next episode. The algorithm below has been proved to be convergence (See
references for the proof
)
Q Learning
Given : State diagram with a goal state (represented by matrix R )
Find : Minimum path from any initial state to the goal state (represented by matrix Q )
Q Learning Algorithm goes as follow
End Do End For 
The above algorithm is used by the agent to learn from experience or training. Each episode is equivalent to one training session. In each training session, the agent explores the environment (represented by Matrix R ), get the reward (or none) until it reach the goal state. The purpose of the training is to enhance the 'brain' of our agent that represented by Q matrix. More training will give better Q matrix that can be used by the agent to move in optimal way. In this case, if the Q matrix has been enhanced, instead of exploring around and go back and forth to the same room, the agent will find the fastest route to the goal state.
Parameter has range value of 0 to 1( ). If is closer to zero, the agent will tend to consider only immediate reward. If is closer to one, the agent will consider future reward with greater weight, willing to delay the reward.
To use the Q matrix, the agent traces the sequence of states, from the initial state until goal state. The algorithm is as simple as finding action that makes maximum Q for current state:
Algorithm to utilize the Q matrix Input: Q matrix, initial state

The algorithm above will return sequence of current state from initial state until goal state.
Tired of ads? Read it off line on any device. Click here to purchase the complete Ebook of this tutorial
<
Previous

Next

Contents
>
Preferable reference for this tutorial is
Teknomo, Kardi. 2005. QLearning by Examples. http://people.revoledu.com/kardi/tutorial/ReinforcementLearning/index.html