2016.06.03 / - - PowerPoint PPT Presentation

▶
2016 06 03
SMART_READER_LITE
LIVE PREVIEW

2016.06.03 / - - PowerPoint PPT Presentation

Apr 22, 2023 273 likes •898 views

2016.06.03 / : - , (2008- ) - (NCsoft) (2006-2008) -

slide-1
SLIDE 1

기계 학습 기반 감정 인식 게임 개발

2016.06.03 홍익대학교 게임학부 / 조교수

강신진

slide-2
SLIDE 2

소개: 강신진

  • 경력
  • 홍익대학교 게임학부, 조교수 (2008-현재)
  • 엔씨소프트(NCsoft) (2006-2008)
  • 소니 컴퓨터 엔터테인먼트 코리아(Sony Computer Entertainment Korea)

(2003-2006)

  • 상용 게임
  • 아이온 (AION), 와일드스타 (WildStar), 블레이드 앤 소울 (Blade and Soul) 외

10여 개의 PC, 모바일, 콘솔 게임 프로젝트 참여

slide-3
SLIDE 3

Evolutionary Game Lab

slide-4
SLIDE 4

EGLAB Research

Emotion Recognition Emotional Model ECA Content

+ +

Multimodal Emotion Recognition Rule-based Model / DQN Embodied Conversational Agent

slide-5
SLIDE 5

Index

  • Machine Learning in Computer Games
  • Affective Computing in Computer Games
  • [ML + Affective Computing] Game Development
slide-6
SLIDE 6

Machine Learning in Commercial Games

slide-7
SLIDE 7

Black and White (Lionhead, 2001)

slide-8
SLIDE 8

Forza Motorsport (Microsoft, 2005)

slide-9
SLIDE 9

F.E.A.R. (Monolith Production, 2005)

slide-10
SLIDE 10

HALO3 (Microsoft, 2007)

slide-11
SLIDE 11

Virtua Fighter Ghost System (SEGA, 2015)

slide-12
SLIDE 12

Tekken 5: Dark Resurrection (Namco, 2016)

slide-13
SLIDE 13

Blade & Soul (NCsoft, 2016)

slide-14
SLIDE 14

ML in Computer Games

Strengths

  • Re-playability
  • Emergent Game Play

Weakness

  • Limitation of CPU resources for AI
  • Low Development Priority
  • Absence of Evaluation Function
  • ML Content is not Fun
  • ML Content is Unpredictable
slide-15
SLIDE 15

Index

  • Machine Learning in Computer Games
  • Affective Computing in Computer Games
  • [ML + Affective Computing] Game Development
slide-16
SLIDE 16

Affective Computing in Commercial Games

slide-17
SLIDE 17

Mindlink (ATARI, 1984)

slide-18
SLIDE 18

Tokimeki Memorial (Konami, 1997)

slide-19
SLIDE 19

Relax to Win Game (McDarby, 2002)

slide-20
SLIDE 20

PlayStation Eyetoy / Microsoft Kinect (2003)

slide-21
SLIDE 21

Mindwave (Neurosky, 2012)

slide-22
SLIDE 22

Affective Computing in Computer Games

Strengths

  • Providing New Experience
  • Market Share Expansion

Weakness

  • Intrusive Additional Hardware
  • Low Recognition Rate of Player’s

Emotion

slide-23
SLIDE 23

Index

  • Machine Learning in Computer Games
  • Affective Computing in Computer Games
  • [ML + Affective Computing] Game Development
slide-24
SLIDE 24

[ML + AC] in Computer Games

slide-25
SLIDE 25

Commercially Successful Games

slide-26
SLIDE 26

[ML + AC] in Computer Games

slide-27
SLIDE 27

[ML + AC] in Computer Games

Strengths

  • Providing New Experience
  • Market Share Expansion

Weakness

  • Intrusive Additional Hardware
  • Low Recognition Rate of Player’s

Emotion Strengths

  • Re-playability
  • Emergent Game Play

Weakness

  • Limitation of CPU resources for AI
  • Low Development Priority
  • Absence of Evaluation Function
  • ML Content is not Fun
  • ML Content is Unpredictable
slide-28
SLIDE 28

[ML + AC] in Computer Games

  • Limitation of CPU resources for AI
  • Low Development Priority
  • Content is not Fun
  • Content is Unpredictable
  • Intrusive Additional Hardware
  • Low Recognition Rate of Player’s Emotion
  • Absence of Evaluation Function
slide-29
SLIDE 29

Solutions

  • Limitation of CPU resources for AI  AI Oriented Content
  • Low Development Priority  Long Term Project
  • Content is not Fun  Appropriate Genre
  • Content is Unpredictable  Non-Competitive Game
  • Intrusive Additional Hardware  Non-Intrusive/Default Hardware
  • Low Recognition Rate of Player’s Emotion  ML with Big Data
  • Absence of Evaluation Function  Robust Emotional Model
slide-30
SLIDE 30

EGLAB Research

Emotion Recognition Emotional Model ECA Content

+ +

Multimodal Emotion Recognition Rule-based Model / DQN Embodied Conversational Agent

slide-31
SLIDE 31

Embodied Conversational Agent (ECA)

GRETA, SEMAINE, HUMAINE (2007) Summer Lesson, BabyX (2016)

slide-32
SLIDE 32

Summer Lesson (Namco, 2015)

slide-33
SLIDE 33

BabyX (Sagar, 2016)

slide-34
SLIDE 34

ECA Core Technologies

Non-Verbal Communication Verbal Communication

slide-35
SLIDE 35

ECA Core Technologies

Non-Verbal Communication Verbal Communication

High-Poly Modeling Facial Expression Gestures / Realistic Animation Eye- Gazing Language Generation Language Understand ing

slide-36
SLIDE 36

Implementation Results

“Love Senor”

slide-37
SLIDE 37

Implementation Results

slide-38
SLIDE 38

Solutions

  • Limitation of CPU resources for AI  AI Oriented Content
  • Low Development Priority  Long Term Project
  • Content is not Fun  Appropriate Genre
  • Content is Unpredictable  Non-Competitive Game
  • Intrusive Additional Hardware  Non-Intrusive/Default Hardware
  • Low Recognition Rate of Player’s Emotion  ML with Big Data
  • Absence of Evaluation Function  Robust Emotional Model
slide-39
SLIDE 39

Hardware for Affective Computing

Intrusive H/W

  • EEG
  • Heart Rate
  • Depth Camera
  • Brain Wave

Non-Intrusive H/W

  • Multimodal Interface
  • Keyboard
  • Mouse
  • Webcam
slide-40
SLIDE 40

EGLAB Research

Emotion Recognition Emotional Model ECA Content

+ +

Multimodal Emotion Recognition Rule-based Model / DQN Embodied Conversational Agent

slide-41
SLIDE 41

Implementation Result

“Emotion Recognition System with Multimodal Interface”

slide-42
SLIDE 42

Implementation Result

“Emotion Tracer Client”

slide-43
SLIDE 43

Solutions

  • Limitation of CPU resources for AI  AI Oriented Content
  • Low Development Priority  Long Term Project
  • Content is not Fun  Appropriate Genre
  • Content is Unpredictable  Non-Competitive Game
  • Intrusive Additional Hardware  Non-Intrusive/Default Hardware
  • Low Recognition Rate of Player’s Emotion  ML with Big Data
  • Absence of Evaluation Function  Robust Emotional Model
slide-44
SLIDE 44

EGLAB Research

Emotion Recognition Emotional Model ECA Content

+ +

Multimodal Emotion Recognition Rule-based Model / DQN Embodied Conversational Agent

slide-45
SLIDE 45

Discrete Emotion Model

  • Ekman’s 6 Universal Emotions
slide-46
SLIDE 46

Dimensional Theory

  • The PAD (Pleasure, Arousal, Dominance) model, Albert Mehrabian and James A. Russell
slide-47
SLIDE 47

Evolutionary Model

  • Plutchik's wheel of emotions
slide-48
SLIDE 48

OCC Model

  • “OCC-model” of emotions (Ortony, Clore & Collins 1988)
slide-49
SLIDE 49

Circuit Model (Neural Darwinism)

slide-50
SLIDE 50

Finite State Machine (FSM)

slide-51
SLIDE 51

Behavior Tree

slide-52
SLIDE 52

Decision Tree

slide-53
SLIDE 53

MHP (Model Human Process)

  • Cognitive Model ( Card, Moran, and Newell )
slide-54
SLIDE 54

ACT-R (Adaptive Control of Thought)

  • Cognitive Architecture (Anderson & Lebiere)
slide-55
SLIDE 55

SOAR (State, Operator And Result)

  • Cognitive Architecture (Laird, 2008; Laird, Newell, & Rosenbloom, 1987; Newell, 1990)
slide-56
SLIDE 56

Deep Q-Learning

slide-57
SLIDE 57

Neural Engineering Framework

  • Nengo: graphical and scripting based software package for simulating large-scale neural systems,
  • SPAUN: 2.5 million simulated neurons
slide-58
SLIDE 58

EGLAB Research

Emotion Recognition Emotional Model ECA Content

+ +

Multimodal Emotion Recognition Rule-based Model / DQN Embodied Conversational Agent

slide-59
SLIDE 59

Future Works: Façade (Mateas & Stern, 2004)

slide-60
SLIDE 60

Future Works: Seaman (SEGA, 1999)

slide-61
SLIDE 61

QnA

경청해 주셔서 감사합니다.

http://www.myeglab.com/