PowerPoint

Chapter 1: Introduction
Modeling
Players
Generating
Content
(Chapter 5)
(Chapter 4)
Game AI Panorama
(Chapter 6)
Frontier
Game AI
Research
(Chapter 7)
AI Methods
(Chapter 2)
Playing
Games
(Chapter 3)
Chapter 2: AI Methods
Power pill eaten
Seek for
pellets
No visible
ghost
Ghost on sight
Chase
Ghosts
Ghosts flashing
Evade
Ghosts
Seek for Pellets
Until ghost on sight
Move
(Priority)
Ghost free
corridor
Pellet Found
Corridor with
pellets
Eat next pellet
Corridor
without pellets
Attack Enemy
Until Health = 0
Spot Enemy
0.5
Mini Gun
Select Weapon
(Probability)
0.2
Pistol
Aim
0.3
Rocket
Launcher
Shoot!
Min
Max
5
5
0
10
10
5
-1
5
3
0
2
0
3
-1
0
0
0
1
1
1
0
1
0
0
1
0
0
0
1
0
1
1
0
1
0
1
0
1
0
0
0
1
1
1
0
1
0
0
1
0
0
0
1
0
1
1
0
0
1
1
0
1
0
1
0
p
0
0
1
1
1
0
1
0
0
1
0
0
1
0
1
0
1
1
0
0
0
1
0
0
1
1
1
0
1
0
0
1
0
0
1
0
1
0
1
1
0
0
0
1
π‘₯2
5
1
4
3
2
6
π‘₯1
Age?
<20
Employed?
no
No Car
>28
[20-28]
Salary?
Economy
yes
City Car
low
Compact
fair
Sports
high
SUV
Ghost
not visible
near
Pellet
Power pill
near
Aim for
pill
far
Evade
ghosts
near
Aim for
pellet
fair
Aim for
pellet
far
Aim for
fruit
π‘₯2
w
π‘₯1
π‘₯1
π‘₯2
π‘₯𝑛
Neuron
𝑀1
𝑀2
𝑔
xβˆ™w +𝑏
𝑀𝑛
𝑏
𝑔(x βˆ™ w + 𝑏)
𝑀14
π‘₯1
π‘₯2
π‘₯3
4
𝑀49
1
𝑀48
5
8
π‘Ž8
6
9
π‘Ž9
2
3
𝑀37
Input Layer
𝑀79
7
Hidden Layer
Output Layer
Selection
Expansion
Tree
Policy
Simulation
Default
Policy
Backpropagation
G
Agent
State (s)
Reward (r)
Environment (e.g. Maze)
Action (Ξ±)
1 𝑀01 𝑀02 𝑀13 𝑀14 𝑀15
𝑀0𝑛
2 𝑀01 𝑀02 𝑀13 𝑀14 𝑀15
𝑀0𝑛
P 𝑀01 𝑀02 𝑀13 𝑀14 𝑀15
𝑀0𝑛
Fitness value 𝑓2
Reward
Convolution
Action
Pooling
Chapter 3: Playing Games
Win
Experience
Player
Non-Player
Motivation
Games as AI testbeds, AI that
challenges players,
Simulation-based testing
Motivation
Playing roles that humans would
not (want to) play, Game balancing
Examples
Examples
Board Game AI (TD-Gammon,
Chinook, Deep Blue, AlphaGo,
Libratus), Jeopardy! (Watson),
StarCraft
Rubber banding
Motivation
Simulation-based testing,
Game demonstrations
Motivation
Believable and human-like agents
Examples
Examples
Game Turing Tests (2kBot
Prize/Mario),Persona Modelling
AI that: acts as an adversary, provides
assistance, is emotively expressive, tells
a story, …
Imperfect
Information
Perfect
Information
Observability
Pac-Man
Atari 2600
Checkers
Chess
Go
Ms Pac-Man
Ludo
Monopoly
Backgammon
Time
Granularity
Super Mario Bros
Halo
StarCraft
Battleship
Scrabble
Poker
Real-Time
Turn-Based
Deterministic
Non-deterministic
Stochasticity
Chapter 4: Generating Content
12%
37%
40%
2%
6%
3%
Art
Manufacturing
Other
Debugging
Marketing
Programming
Content
Method
Content Type
Necessary Content
Stochastic
Determinism
Deterministic
Non-Controllable
Controllability
Controllable
Constructive
Iterativity
Generate-and-test
Autonomous
Autonomy
Mixed-Initiative
Experience-Agnostic
Experience
Experience-Driven
Role
Optional Content
Generation 0
Midpoint displacement
Generation 1
Generation 2
Generation 3
Final Generation
Initialize Corner Values
Perform Diamond Step
Perform Square Step
Perform Diamond Step
Perform Square Step
Experience
Agnostic
Experience
Driven
Player
(Experience)
Super Mario Bros
(Pedersen et al., 2010)
Sonancia (Lopes et al.,
2015)
SpeedTree (IDV, 2002)
StarCraft Maps
(Togelius et al., 2013)
Autonomous
Sentient Sketchbook
(Liapis et al., 2013)
Garden of Eden Creation
Kit (Bethesda, 2009)
Ropossum (Shaker et al.,
2013)
Tanagra (Smith et al.,
2010)
Mixed-Initiative
Designer
(Initiative)
Chapter 5: Modeling Players
Input
Computational (Player) Model
Output
Gameplay
Model-Based [Top-Down]
Objective
(Psychology, Cognitive Science, Game
Studies, …)
Free Response vs.
Forced Response
Context
Player Profile
First Person vs.
Third Person
Discrete vs.
Continuous
Numerical (Interval)
Regression
Time-Discrete vs.
Time-Continuous
Nominal (Classes)
Classification
Pre vs. During vs. Post
Ordinal (Ranks)
Preference Learning
No Output
Unsupervised Learning
(Clustering, Frequent Pattern Mining)
Model Free [Bottom-Up]
(Data Science, Machine Learning)
Activation ( + )
Arousal
Fear
Excitement
Anger
Frustration
Happiness
Valence
Pleasant ( + )
Unpleasant ( - )
Sadness
Relaxation
Boredom
Deactivation ( - )
Tiredness
Challenge
Flow
Channel
Anxiety
Boredom
Skills
Nearest
Monster
Nearest
Treasure
Monster
Nearest
Treasure
Safe
Treasure
Nearest
Portion
Safe
Nearest
Portion
Treasure
Safe
Portion
Exit
Portion
Safe
Hit
Points
Exit Safe
Exit
Exit Safe
Chapter 6: Game AI Panorama
DO (Process)
WHAT (Context)
FOR WHO (End User)
Designer
Model
Content
Player
AI Researcher
Generate
Behavior
Producer /
Publisher
GAME AI AREA
Generate Content (Assisted)
Model Players (Behavior, Experience)
Generate Content (Autonomously)
Play Games
(Win [NPC], Experience [NPC])
Play Games (Win [PC], Experience
[PC])
Model Players (Behavior)
Player
Interaction
Model Players
Experience
Behavior
Game
Generate Content
Autonomously
Content
NPCs
Play Games [as NPC]
Win
Experience
Play Games (as PC or NPC)
To Win
Model Players
Experience
Behavior
For the (Game)
Experience
Generate Content
Assisted
Autonomously
Play Games (as PC or NPC)
To Win
Model Players
Experience
Behavior
For the (Game)
Experience
Generate Content
Assisted
Autonomously
Play Games (as PC or NPC)
To Win
Model Players
Experience
Behavior
For the (Game)
Experience
Generate Content
Assisted
Autonomously
Play Games (as PC or NPC)
To Win
Model Players
Experience
Behavior
For the (Game)
Experience
Generate Content
Assisted
Autonomously
Play Games (as PC or NPC)
To Win
Model Players
Experience
Behavior
For the (Game)
Experience
Generate Content
Assisted
Autonomously