44
Automated Assessment for Adaptive Learning of Complex Tasks Alan Koenig Markus Iseli Allen Munro CCT

Automated Assessment for Adaptive Learning of Complex Tasks

  • Upload
    xanto

  • View
    52

  • Download
    0

Embed Size (px)

DESCRIPTION

Automated Assessment for Adaptive Learning of Complex Tasks. Alan Koenig Markus Iseli Allen Munro. CCT. Characteristics of Cognitively Complex Tasks. Often require multiple, non-trivial steps to complete Performance can be highly variable Often include interdependent task features - PowerPoint PPT Presentation

Citation preview

Page 1: Automated Assessment for  Adaptive Learning of Complex Tasks

Automated Assessment for Adaptive Learning of Complex Tasks

Alan KoenigMarkus IseliAllen Munro

CCT

Page 2: Automated Assessment for  Adaptive Learning of Complex Tasks

Characteristics of Cognitively Complex Tasks Often require multiple, non-trivial steps to complete

Performance can be highly variable

Often include interdependent task features

Often well suited to representation in games & simulations– Multiple scenarios

– Low costExamples:

Tactical Planning Simulation Damage Control Simulation

– Low risk– But… assessment can be difficult

Page 3: Automated Assessment for  Adaptive Learning of Complex Tasks

Anti-Submarine Warfare

Anti-submarine warfare (ASW) is a branch of naval warfare that uses surface warships, aircraft, or other submarines to find, track and deter, damage, or destroy enemy submarines.

– Carried out by teams of people within and across ships

– Requires synthesis of large quantities of information in short periods of time

– Heavy emphasis on situation awareness and decision making skills (especially among TAO’s)

Page 4: Automated Assessment for  Adaptive Learning of Complex Tasks

Assessment Methodology Overview

Page 5: Automated Assessment for  Adaptive Learning of Complex Tasks

ASW Ontology Construction

ExpertsRules /Doctrine

Procedures /Processes

Facts /Observations

DOMAIN

Page 6: Automated Assessment for  Adaptive Learning of Complex Tasks

Ontology Excerpt

Page 7: Automated Assessment for  Adaptive Learning of Complex Tasks

Bayesian Network Development

OntologyNodes

Directionalityof Links

Observables /SME Input

Strength of Relationships

Characterizing the nature of relationships is essential

Observable Actions

Latent VariablesSonar

Planning

PassiveSonar Planning

ActiveSonar

Planning

UseShipPassiveSonar DropPassiveSonobuoys

DropActiveSonobuoys UseDippingSonar UseShipActiveSonar

Page 8: Automated Assessment for  Adaptive Learning of Complex Tasks

ASW Skilled

Torpedo CounterMeasures

MEUProtectionSub Hunting

Plan for Opposition

Intell on Opposition

Opposition Estimation

ShowMissionInfo ShowBriefing

EstLLSubSpeed

EstLLTorpedoRange

0.5

SonarPlanning

TDA AvoidManeuvers

PassiveSonar Use

ActiveSonar Skill

AlterCourseTo AvoidFast

AlterCourseTo AvoidSlow

UseShipPassiveSonar DropPassiveSonobuoys

DropActiveSonobuoys UseDippingSonar UseShipActiveSonar

TDA Avoidance

DontEverEnterTDALimLines

Estimation

EstDatumSubSpeed EstDatumTorpedoRange

Datum Estimation

TorpedoDeception

TorpedoEscape

NixieDropped MICDropped

TorpedoEvasion

TorpedoCounter

Bayesian Network for TDA Avoidance

Page 9: Automated Assessment for  Adaptive Learning of Complex Tasks

ASW Skilled

Torpedo CounterMeasures

MEUProtectionSub Hunting

Plan for Opposition

Intell on Opposition

Opposition Estimation

0.5

TDA AvoidManeuvers

DropActiveSonobuoys UseDippingSonar UseShipActiveSonar

TDA Avoidance

LimLines Estimation

Datum Estimation

TorpedoDeception

TorpedoEscape

Bayesian Network for TDA Avoidance

ShowMissionInfo ShowBriefing

EstLLSubSpeed

EstLLTorpedoRange

SonarPlanning

PassiveSonar Use

ActiveSonar Skill

AlterCourseTo AvoidFast

AlterCourseTo AvoidSlow

UseShipPassiveSonar DropPassiveSonobuoys

DontEverEnterTDA

EstDatumSubSpeed EstDatumTorpedoRange

NixieDropped MICDropped

TorpedoEvasion

TorpedoCounter

Page 10: Automated Assessment for  Adaptive Learning of Complex Tasks

Conditions for TDA Avoidance to be considered by Bayesian network

Page 11: Automated Assessment for  Adaptive Learning of Complex Tasks

Observable Actions

Latent Variables

The Advantage of Bayesian Networks

SonarPlanning

PassiveSonar Planning

ActiveSonar

Planning

UseShipPassiveSonar DropPassiveSonobuoys

DropActiveSonobuoys UseDippingSonar UseShipActiveSonar

Probability of mastery of the latent variables is inferred from the scored observable actions

P(ASP): Probability of mastery of concept Active Sonar Planning

P(UDS | ASP): Conditional probability. Probability of mastery of concept UseDippingSonar, given information about mastery of concept Active Sonar Planning

Example: Determining probabilities

P(UDS):Probability of mastery of concept UseDippingSonar

P(ASP) x P(UDS | ASP)

P(UDS)P(ASP | UDS) =

Page 12: Automated Assessment for  Adaptive Learning of Complex Tasks

Real-Time Formative Assessment

•Provide performance feedback•Provide practice / resources•Add / change tasks•Add / change affordances

FORMATIVE ASSESSMENT

Simulation(Sandbox)

ADAPTIVE OPTIONS

Observable Actions & Events

CAA(with Bayesian Network)

Page 13: Automated Assessment for  Adaptive Learning of Complex Tasks
Page 14: Automated Assessment for  Adaptive Learning of Complex Tasks

Check the Briefing—Good!

1st PROBLEM: Chinese Kilos Threat Problem

00:00:00 .k.Announce. ShowBriefing 0.95

The fact that we’re up against Kilos implies a silent sub speed of 6 Kt.and a torpedo range of 12,000 yds.

Page 15: Automated Assessment for  Adaptive Learning of Complex Tasks

Failed to Correct Estimates—Bad!

1st PROBLEM: Chinese Kilos Threat Problem

07:26:32 .k.Announce. EstSubSpeed 0.207:26:32 .k.Announce. EstTorpedoRange 0.2

The estimated sub speed was left at 3 Kt. So the limit lines were not accurate, relative to the briefing.

Scored at end, because these values were never changed.

Page 16: Automated Assessment for  Adaptive Learning of Complex Tasks

Failed to Correct Estimates—Bad!

1st PROBLEM: Chinese Kilos Threat Problem

07:26:32 .k.Announce. DontEverEnterTDA 0.1

Uh-oh! The MEU entered the TDZ.

Page 17: Automated Assessment for  Adaptive Learning of Complex Tasks

17

Problem 1

Page 18: Automated Assessment for  Adaptive Learning of Complex Tasks

18

Problem 1

Page 19: Automated Assessment for  Adaptive Learning of Complex Tasks

19

Problem 1

Page 20: Automated Assessment for  Adaptive Learning of Complex Tasks

20

Problem 1

Page 21: Automated Assessment for  Adaptive Learning of Complex Tasks

21

Problem 1

Page 22: Automated Assessment for  Adaptive Learning of Complex Tasks

Check the Briefing—Good!

2nd PROBLEM: Threat West of Gibraltar

This time the mission brief gives specific estimates of hostile sub speed and torpedo range.

00:00:00 .k.Announce. ShowBriefing 0.95

Page 23: Automated Assessment for  Adaptive Learning of Complex Tasks

Datum Estimates2nd PROBLEM: Threat West of Gibraltar

The datum estimates forsub speed and torpedorange don’t match the briefing.

Page 24: Automated Assessment for  Adaptive Learning of Complex Tasks

Datum Estimates Corrected—Good!

2nd PROBLEM: Threat West of Gibraltar

Set to the correct values.

01:14:03 .k.Announce. EstSubSpeed 0.9501:14:03 .k.Announce. EstTorpedoRange 0.95 Scored at end.

Page 25: Automated Assessment for  Adaptive Learning of Complex Tasks

Check Limit Lines2nd PROBLEM: Threat West of Gibraltar

Perhaps not a wide enough berth…

Page 26: Automated Assessment for  Adaptive Learning of Complex Tasks

Plot a New Course2nd PROBLEM: Threat West of Gibraltar

A safer course?

Page 27: Automated Assessment for  Adaptive Learning of Complex Tasks

Check Limit Lines Again2nd PROBLEM: Threat West of Gibraltar

Looks safer

Page 28: Automated Assessment for  Adaptive Learning of Complex Tasks

Successful Avoidance2nd PROBLEM: Threat West of Gibraltar

Success!

01:14:03 .k.Announce. DontEverEnterTDA 0.95

Page 29: Automated Assessment for  Adaptive Learning of Complex Tasks

29

Problem 2

Page 30: Automated Assessment for  Adaptive Learning of Complex Tasks

30

Problem 2

Page 31: Automated Assessment for  Adaptive Learning of Complex Tasks

31

Problem 2

Page 32: Automated Assessment for  Adaptive Learning of Complex Tasks

32

Problem 2

Page 33: Automated Assessment for  Adaptive Learning of Complex Tasks

33

1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Problem 1Problem 2

1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Problem 1Problem 2

Plan for Opposition

TDA Avoidance

Sub speed: Wrong estimate

Entered TDA

Prob

abili

ty o

f Mas

tery

Prob

abili

ty o

f Mas

tery

time

time

Page 34: Automated Assessment for  Adaptive Learning of Complex Tasks

ASW Skilled

Torpedo CounterM.

MEUProtectionSub Hunting

Plan for Opposition

Intelligence on Opposition

Opposition Estimation

GetMissionInfo GetBriefing EstSubSpeed EstTorpedoRange

SonarPlanning

TDA AvoidManeuvers

SenseQuietly

SenseLoudly

ManeuverFastManeuverSlow

ShipsPassivSonar PassiveSonobuoys ActiveSonobuoys DippingSonar ShipsActivSonar

TDA Avoidance

B20: score 0.95D20: score 0.6F20: score 0.1

B16: score 0.95D16: score 0.7F16: score 0.05 B5: score 0.9

D5: score 0.7B6: score 0.95D6: score 0.7

B11: score 0.95D11: score 0.8F11: score 0.2

B12: score 0.95D12: score 0.8F12: score 0.2

B36: score 0.9E36: score 0.3F36: score 0.2

B40: score 0.95C40: score 0.4E40: score 0.3

C44: score 0.9D44: score 0.6F44: score 0.05

B28: score 0.95D28: score 0.6

B32: score 0.95E32: score 0.3

DontEnterTDAIfPos

B24: score 0.95C24: score 0.9F24: score 0.1

TDA avoidanceBayesian network sub-net

Page 35: Automated Assessment for  Adaptive Learning of Complex Tasks

Initial Evaluation Feedback

CAA Debugging OutputInstructional Feedback High Level Assessments

Page 36: Automated Assessment for  Adaptive Learning of Complex Tasks

First Successful Problem

Page 37: Automated Assessment for  Adaptive Learning of Complex Tasks

Into the Next Problem

Page 38: Automated Assessment for  Adaptive Learning of Complex Tasks

Adaptive Problem Selection

A new problem is advised.

Page 39: Automated Assessment for  Adaptive Learning of Complex Tasks

Loading the Problem

The advised problem is selected

Page 40: Automated Assessment for  Adaptive Learning of Complex Tasks

Initial Actions in the New Problem

Page 41: Automated Assessment for  Adaptive Learning of Complex Tasks

Preparing to Find the Sub

Launching a helicopter

Page 42: Automated Assessment for  Adaptive Learning of Complex Tasks

Detecting the Sub

Sonobuoys detect sub

Page 43: Automated Assessment for  Adaptive Learning of Complex Tasks

Recording of Performances & Assessments

Simulation(Sandbox)

Log Files(Actions and Events)

These performance recordscan be replayed.

Report Generation(CAA Monitor)

Assessment History(Per Student)

Report

CAA

Page 44: Automated Assessment for  Adaptive Learning of Complex Tasks

Questions / Comments?

CCT