+ All Categories
Home > Documents > Machine Learning Methods for Parameter Acquisition in a Human ...

Machine Learning Methods for Parameter Acquisition in a Human ...

Date post: 18-Jul-2015
Category:
Upload: butest
View: 203 times
Download: 1 times
Share this document with a friend
Popular Tags:
21
Machine Learning Approaches to Cognitive Parameter Acquisition Terran Lane University of New Mexico [email protected] Chris Forsythe, Patrick Xavier Sandia National Labs {jcforsy,pgxavie}@sandia.gov
Transcript
Page 1: Machine Learning Methods for Parameter Acquisition in a Human ...

Machine Learning Approaches to Cognitive Parameter Acquisition

Terran LaneUniversity of New Mexico

[email protected]

Chris Forsythe, Patrick Xavier

Sandia National Labs{jcforsy,pgxavie}@sandia.gov

Page 2: Machine Learning Methods for Parameter Acquisition in a Human ...

Sandia’s Cognitive Modeling Framework

Computational models of human decision-makers

Models attention, perceptual cues, situational awareness, decision making

Based on oscillatory models of activation Spreading activation networks and feedback

loops between functional elements Applications -- data analysis, security, tutoring… Bottleneck: models hand-built/tuned

Expensive and slow!

Page 3: Machine Learning Methods for Parameter Acquisition in a Human ...

The Big PictureW

orld

Cue0 ω0

Cue1 ω1

CueN ωN

Situation0

Situation1

SituationM

Actio

ns/

Dec

isio

ns

θ01

θ10

θN1

θNM

Φ

Page 4: Machine Learning Methods for Parameter Acquisition in a Human ...

Automated Model Acquisition High predictive accuracy

87% correct prediction of operator’s interpretation of scenario (incl. relevance)

91% correct in recognizing situation only Insights into operator decision-making process Models are task & user specific

Only 26% overlap between users Large effort in building and tuning models

Project goal: (semi-)automate acquisition of parameters, network topologies, etc.

Prediction accuracy secondary concern

Page 5: Machine Learning Methods for Parameter Acquisition in a Human ...

Roles for Machine Learning

Parameter acquisition Interconnection weights Activation levels Oscillator frequencies

Network topologies Inter-cue spreading activation network Cue <-> situation relations Feedbacks

Cues and situation identification

Page 6: Machine Learning Methods for Parameter Acquisition in a Human ...

Parameter AcquisitionW

orld

Cue0 ω0

Cue1 ω1

CueN ωN

Situation0

Situation1

SituationM

Actio

ns/

Dec

isio

ns

θ01

θ10

θN1

θNM

Φ

Page 7: Machine Learning Methods for Parameter Acquisition in a Human ...

Parameter Acquisition: Issues

Superficially supervised learning Observe features/cues and operator actions;

induce params (find Θ s.t. fΘ:CA) Similar to ANN backprop, EM, etc. Many effective, well understood techniques

Problem: not just high-likelihood params Actually want params used by human operator Much harder – observable stimuli don’t directly

reflect operator’s internal state Cognitive plausibility constraint

Page 8: Machine Learning Methods for Parameter Acquisition in a Human ...

Parameter Acquisition: Approaches

Additional instrumentation Measure characteristics of operator Biometrics – eye tracking, MEG, etc. Expensive, not widespread Maybe not informative to params anyway

Utility elicitation techniques Software queries user about why decisions

were made / state of attention Picks questions to maximally improve model Emulates expert knowledge engineer

Page 9: Machine Learning Methods for Parameter Acquisition in a Human ...

Network Topology InductionW

orld

Cue0 ω0

Cue1 ω1

CueN ωN

Situation0

Situation1

SituationM

Actio

ns/

Dec

isio

ns

θ01

θ10

θN1

θNM

Φ

Page 10: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Issues

Find structure of interconnections between variables (I.e., cues, situations)

Much harder than parameter acquisition Formally, maximum likelihood/MAP search

through all possible networks

Page 11: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Issues

Find structure of interconnections between variables (I.e., cues, situations)

Much harder than parameter acquisition Formally, maximum likelihood/MAP search

through all possible networks

L=137

Page 12: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Issues

Find structure of interconnections between variables (I.e., cues, situations)

Much harder than parameter acquisition Formally, maximum likelihood/MAP search

through all possible networks

L=137 L=238

Page 13: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Issues

Find structure of interconnections between variables (I.e., cues, situations)

Much harder than parameter acquisition Formally, maximum likelihood/MAP search

through all possible networks

L=137 L=238 L=493

Page 14: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Issues

Find structure of interconnections between variables (I.e., cues, situations)

Much harder than parameter acquisition Formally, maximum likelihood/MAP search

through all possible networks

L=137 L=238 L=493 L=318

Page 15: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Issues

Find structure of interconnections between variables (I.e., cues, situations)

Much harder than parameter acquisition Formally, maximum likelihood/MAP search

through all possible networks

L=137 L=238 L=493 L=318

Page 16: Machine Learning Methods for Parameter Acquisition in a Human ...

Topology Induction: Approaches

Principles of structure search well understood Gradient ascent, annealing, genetic search,

constrained search, etc. Difficult in practice

Computationally intractable Resulting models very sensitive to data Spurious likelihood spikes low confidence

models Compounded by cognitive plausibility constraint Can get leverage from cognitive plausibility,

though

Page 17: Machine Learning Methods for Parameter Acquisition in a Human ...

Cue and Situation IdentificationW

orld

Cue0 ω0

Cue1 ω1

CueN ωN

Situation0

Situation1

SituationM

Actio

ns/

Dec

isio

ns

θ01

θ10

θN1

θNM

Φ

Page 18: Machine Learning Methods for Parameter Acquisition in a Human ...

Cue and Situation Identification: Issues

Discern cues and whole environmental situations employed by user

Related to constructive feature induction, nonlinear projection identification, relational learning, etc.

Search across all possible nodes/relations

N=2 N=3

Page 19: Machine Learning Methods for Parameter Acquisition in a Human ...

Cue and Situations: Approaches

Cutting-edge ML problem Direct elicitation is probably most promising

approach Formulating search space/uncertainty reduction

not straightforward Even user interface is difficult (naming synthetic

nodes/relations)

Page 20: Machine Learning Methods for Parameter Acquisition in a Human ...

Conclusions

Decrease time/effort/cost to construct and tune cognitive model

Constrained to correspond to human’s internal model Both bane and boon to automated model

construction Insights into operator’s mental state/decision-

making process Requires/drives novel ML algorithms Future work: all of it…

Page 21: Machine Learning Methods for Parameter Acquisition in a Human ...

Questions?


Recommended