emotional agent : a modeling and an application

35
Emotional agent : A modeling and an application Khulood et al. Information and Software Technology vol. 49, pp. 695-716, 2007 장장장

Upload: yaron

Post on 25-Feb-2016

26 views

Category:

Documents


1 download

DESCRIPTION

Emotional agent : A modeling and an application. Khulood et al. Information and Software Technology vol. 49, pp. 695-716, 2007 장수형. Introduction. Emotion Essential part of human Influence how we adapt, learn, behavior, communicate with others - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Emotional agent  : A modeling and an application

Emotional agent : A modeling and an application

Khulood et al. Information and Software Technology

vol. 49, pp. 695-716, 2007

장수형

Page 2: Emotional agent  : A modeling and an application

Introduction

• Emotion– Essential part of human– Influence how we adapt, learn, behavior, communicate

with others– Important and active role in the human decision-making

process– Use artificial agent as test bed– Exploit some of the roles that emotions play in biological

system• Develop mechanisms and tools to ground and enhance autonomy

• Definition of emotion– Occurring when the cognitive, physiological, and motor/

expressive components are usually more or less dissoci-ated in serving separate functions

– Psychological states

Page 3: Emotional agent  : A modeling and an application

Introduction

• The different affective states– Emotion

• Angry, sad, joyful, fearful, ashamed, proud, elated, desperate– Mood

• Cheerful, gloomy, irritable, listless, depressed, buoyant– Preferences/Attitudes

• Liking, loving, hating, valuing, desiring• Emotion History

– Psychology, Neurology, Philosophy, Cognitive science– ‘The Emotional Brain’ – LeDoux

• Emotional process in the brain– Terms of desires and expectations

Page 4: Emotional agent  : A modeling and an application

Emotion

• Role of emotion in nature– Serve several crucial roles in animals and human alike– Provide a basic evaluation in terms of hedonic values– Cause the organism to be attracted to what it likes and to

avoid what it does not like• Fear-anger system may generate fight or flight behavior

– Influence direct cognitive process, process strategies– Play an important role in social contexts

• Raging from signaling emotional state through facial expressions and gesture

Page 5: Emotional agent  : A modeling and an application

Emotion

• Role of emotions in artificial agents– Action selection– Adaptation– Social regulation– Sensory integration– Alarm mechanisms– Motivation– Goal management– Learning– Attention focus– Memory control– Strategic processing– Self-model

Page 6: Emotional agent  : A modeling and an application

Emotion

• Emotion cognitive appraisal– Complex , dynamic, varying both episodically and longi-

tudinally– A negative event can trigger an emotional response

• Dissipate within a short time

Page 7: Emotional agent  : A modeling and an application
Page 8: Emotional agent  : A modeling and an application

Emotion

• OCC-model– Vague, vary, difficult to tell apart

• A lot of confusion– Basic emotions, reduce everything– Consider the best categorization of emotion– World as divided in three different categories

• Events, agents, objects– Categories down to five distinct positive and five nega-

tive

Page 9: Emotional agent  : A modeling and an application

Emotional agent modeling

• Symbolic approach one• Model the behavior of two agents and objects living in a simulated world

• RIA(Regular Intelligent Agent), EIA(Emotional Intelli-gent Agent)

– Same object– Main goal to achieve certain activities

• Agent’s global variables and states can be monitored through the simulation using graph, plot, report

Page 10: Emotional agent  : A modeling and an application

Emotional agent modeling

• Model hypothesis– The mission ‘To bring life’ to several application

• Information, transaction, education, tutoring, business, entertainment and e-commerce

– Develop artificial mechanisms• Can play the role emotion plays in natural life• Artificial emotions

– NetLogo• Extensions of the LOGO• Control many agent on the screen• http://ccl.northwestern.edu/netlogo.• An agent modeling environment• Well suited for modeling complex system• Hundreds or thousands of independent ‘agents’

Page 11: Emotional agent  : A modeling and an application

Netlogo

• System– Run on MacOS, Windows, Linux, et al.– Model can be saved as applets to be embedded in web

• Language– Fully programmable– Simple language structure– Language is Logo – Unlimited number of agent and variable– Integer and double precision floating-point math– Runs are exactly reproducible cross platform

• Environment– Graphics display supports turtle shapes and size, exact

turtle positions, and turtle and patch label– Interface builder with buttons, sliders, switches, choices,

monitors, and text boxes

Page 12: Emotional agent  : A modeling and an application

Orphanage scenario

• Require the develop of a Netlogo environment provid-ing a set of behavior rules

– Be used for the simulation of agent behavior– Emotional agent behavior toolkit

• The Orphanage Care Problem– Two agents

• To achieve agent’s main goal– Agent should go to Orphanage– Taking care of the Orphanage depends on the agent

working capacity– Taking care of the Orphanage depends on the agent earn-

ing level

Page 13: Emotional agent  : A modeling and an application

Orphanage scenario

• To preserve earning level from decline to zero– Agent should go to work to make some money

• Working capacity can be improved at the Academy– Agent should go to the Academy

• Improve its knowledge, hence, earning salary– Agent must pay fee – Working capacity do not decay over time

• Agent need to raise its social capacity– Agent should go a social place such as club, restaurant,

mall, party– Going their need expense which costs money– Social capacity do not decay over time

Page 14: Emotional agent  : A modeling and an application

Emotion on Orphanage

• Important roles at the control-level of agents behav-ior

– Lead a reflexive reaction– Support the goal and motivation of an agent– Can create new motivations– Operate a the lever of control of agent architecture– Behavior of the agent will improve– Agent can generate emotion signal, evaluate and as-

sesses events• Integration of Agent Goal, Personality, behavior

– Netlogo can be varying initial conditions

Page 15: Emotional agent  : A modeling and an application

Agent attributes

Page 16: Emotional agent  : A modeling and an application

Agent attributes

• Agent attributes

Page 17: Emotional agent  : A modeling and an application

Simulation of objects

Page 18: Emotional agent  : A modeling and an application

• The Orphanage• Job• Club• Academy• Main-Goal Status• Agent’s Performance Measure

– Social capacity– Working capacity– Earning level

• Make sure– Their orphanage status does not decay completely– They do not run out of money

Page 19: Emotional agent  : A modeling and an application

General Concept

Page 20: Emotional agent  : A modeling and an application

Agent goals

• Main goal– Take care of the Orphanage and as good as they can

• Sub goal

Page 21: Emotional agent  : A modeling and an application

Emotions in the simulation model

• Emotions come into play just after event perception– Compare to its goals and standards– Attitudes are also…– Result

• Value of Event-based emotion, attribution emotions and attraction emotions

– The appraisal/evaluation mechanism • OCC-theory

– Rise to emotions• Less complex than the full human emotion spectrum• Play a meaningful role in the mental processes

Page 22: Emotional agent  : A modeling and an application

Emotions in the simulation model

• Emotion parameters– Event-based emotions

• Be influenced by the level of Orphanage-Status, earning level, social, work capacity

– Attribution emotions• Be affected by the measure of other agents(work, learn or socialize

with)– Attraction emotions

• Works the same as attribution emotion• With liking/disliking of object

• Agent behavior– ‘Behaviors with perception’ concept

• Depend upon the current state of environment(perception)• State of the world, other agent

Page 23: Emotional agent  : A modeling and an application

RIA

• Thinking process(RIA)– Receives its initial states of it memory, parameter– Perceive behavioral Environment– Goal filtering– Behavior/action

Page 24: Emotional agent  : A modeling and an application

• Perception(behavior environment scanning)– Scans the environment and gathers data about object

features– Assign priorities to the goal action

• The orphanage status level• The earning level

– Perception rules• Checks money level and give certain priority for job

Page 25: Emotional agent  : A modeling and an application

• Reasoning(simple inference)– If there are two goal-actions with equal priority– Decide which action is important

Page 26: Emotional agent  : A modeling and an application

• Execution– With highest priority

• Going-to-job, learning-at-academy, socialize-at-clue– Result can be noticed

• Orphanage status level• Working/social capacity• Earning level

Page 27: Emotional agent  : A modeling and an application

EIA

• Thinking process(EIA)– Receive its initial states of its memory and other initial

parameter– Perceive the behavioral environment– Appraisal for situation is performed– Emotion generation– Emotion normalization– Personality influence– Goal Filtering– Action

Page 28: Emotional agent  : A modeling and an application

EIA• Evaluate agent’s attitudes

Page 29: Emotional agent  : A modeling and an application

EIA

Page 30: Emotional agent  : A modeling and an application

Simulation

• Test– 2000 iteration of the simulation

Page 31: Emotional agent  : A modeling and an application

Result

• Run 10 times

Page 32: Emotional agent  : A modeling and an application

Result

Page 33: Emotional agent  : A modeling and an application

Result

Page 34: Emotional agent  : A modeling and an application

conclusions

• Artificial emotions can be used in different ways to influence decision-making

• Orphanage Care Problem• Emotions can be successfully modeled in agent• Outperform it non-emotional counterpart• EIA can be used in decision making in dynamic system model

Page 35: Emotional agent  : A modeling and an application

E.N.D