This document provides an introduction to reinforcement learning. It defines reinforcement learning and distinguishes it from supervised and unsupervised learning. It presents a mathematical formula for calculating rewards received at each state and provides two examples of how reinforcement learning could be used: determining optimal ad placement on a web page to maximize revenue, and controlling a walking robot to reach a target destination. It also briefly mentions the epsilon greedy algorithm.

1. INTRODUCTIO
N
TO
Dr. Tri Basuki Kurniawan

2. Definition
•
•

3. RL vs Supervised and
Unsupervised Learning
1.
2.
3.
4.

4. Some Math Stuff
𝐺𝑡 = 𝑘=0
∞
𝛾𝑘
𝑅𝑡+𝑘+1𝑤ℎ𝑒𝑟𝑒 𝛾 𝜖[0,1)
𝑅𝑡+1 + 𝛾𝑅𝑡+2 + 𝛾2𝑅𝑡+3 …
Sigma (sum up)
Discount rate
Rewards receive at each state
Expanded from of the Equation

5. Example of Use Case
Agent: The program making
decision on how many ads are
appropriate for a page.
Environment: The web page.
Action: One of three: (1)
putting another ad on the
page; (2) dropping an ad from
the page; (3) neither adding
nor removing .
Reward: Positive when
revenue increase; negative
when revenue drops.
Determining the Placement of Ads on a Web Page

6. Example of Use Case
Controlling a Walking Robot
AGENT: THE PROGRAM
CONTROLLING A WALKING
ROBOT
ENVIRONMENT: THE REAL
WORLD.
ACTION: ONE OF FOUR MOVE: (1)
FORWARD; (2) BACKWARD; (3)
LEFT AND (4) RIGHT.
REWARD: POSITIVE WHEN IT
APPROACHES THE TARGET
DESTINATION; NEGATIVE WHEN IT
WASTES TIME, GOES IN THE
WRONG DIRECTION OR FALL
DOWN.

7. Epsilon Greedy Algorithm

8. THANK YOU
Dr. Tri Basuki Kurniawan