Modeling Discrete Interventional Data using Directed Cyclic Graphical Models

English
84 pages
Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus

Description

Modeling Discrete Interventional Data using Directed Cyclic Graphical Models Mark Schmidt and Kevin Murphy Department of Computer Science University of British Columbia June 21, 2009

  • intracellular multivariate flow

  • discrete interventional

  • murphy modeling

  • introduction interventional

  • potential model

  • data using


Sujets

Informations

Publié par
Nombre de lectures 27
Langue English
Poids de l'ouvrage 1 Mo
Signaler un problème

ModelingDiscreteInterventionalDatausing

DirectedCyclicGraphicalModels

MarkSchmidtandKevinMurphy

DepartmentofComputerScience

UniversityofBritishColumbia

June21,2009

siDgniledoMyhpruM.KdnatdimhcS.MnoitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitExperiments

n4

eImplementation

t3

o2

PInterventionalPotentialModel

lIntroduction
Motivation
OurContribution

a1

nOutline

oitnevretnInoitcudortnIsledoMGCDgnisuataDlanoitnevretnIeterc
sledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.MnoitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIIfI
set
mywatchsoitsays11:55,itdoesn’thelp

IfI
see
thatmywatchsays11:55,thenit’salmostlunchtime

Thedifferencebetween
conditioningbyobservation
and
conditioningbyintervention
inthe‘hungryatwork’problem:

collectsbothobservationaland
interventional
data.

collectsalargenumberofsamples

simultaneouslymeasuresmultiplemolecules

Recently,Sachsetal.[2005]analyzedanintracellularmultivariate
flowcytometrydatasetthat:

MotivatingProblem:ModelingBiologicalNetworks

noitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIIfI
set
mywatchsoitsays11:55,itdoesn’thelp

IfI
see
thatmywatchsays11:55,thenit’salmostlunchtime

Thedifferencebetween
conditioningbyobservation
and
conditioningbyintervention
inthe‘hungryatwork’problem:

collectsbothobservationaland
interventional
data.

collectsalargenumberofsamples

simultaneouslymeasuresmultiplemolecules

Recently,Sachsetal.[2005]analyzedanintracellularmultivariate
flowcytometrydatasetthat:

MotivatingProblem:ModelingBiologicalNetworks

sledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.M
sledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.MnoitubirtnoCruOnoitavitoMstnemiIfI
set
mywatchsoitsays11:55,itdoesn’thelp

rIfI
see
thatmywatchsays11:55,thenit’salmostlunchtime

eThedifferencebetween
conditioningbyobservation
and
conditioningbyintervention
inthe‘hungryatwork’problem:

pcollectsbothobservationaland
interventional
data.

xcollectsalargenumberofsamples

Esimultaneouslymeasuresmultiplemolecules

nRecently,Sachsetal.[2005]analyzedanintracellularmultivariate
flowcytometrydatasetthat:

oMotivatingProblem:ModelingBiologicalNetworks

itatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnI
sledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.MnoitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIIfI
set
mywatchsoitsays11:55,itdoesn’thelp

IfI
see
thatmywatchsays11:55,thenit’salmostlunchtime

Thedifferencebetween
conditioningbyobservation
and
conditioningbyintervention
inthe‘hungryatwork’problem:

collectsbothobservationaland
interventional
data.

collectsalargenumberofsamples

simultaneouslymeasuresmultiplemolecules

Recently,Sachsetal.[2005]analyzedanintracellularmultivariate
flowcytometrydatasetthat:

MotivatingProblem:ModelingBiologicalNetworks

noitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIsledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.MIfI
set
mywatchsoitsays11:55,itdoesn’thelp

IfI
see
thatmywatchsays11:55,thenit’salmostlunchtime

Thedifferencebetween
conditioningbyobservation
and
conditioningbyintervention
inthe‘hungryatwork’problem:

collectsbothobservationaland
interventional
data.

collectsalargenumberofsamples

simultaneouslymeasuresmultiplemolecules

Recently,Sachsetal.[2005]analyzedanintracellularmultivariate
flowcytometrydatasetthat:

MotivatingProblem:ModelingBiologicalNetworks

irepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIMotivating

Problem:

Networks

Modeling

Biological

sledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.MnoitubirtnoCruOnoitavitoMstnem
noitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIsledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.MButDAGsdonotallowthemodeltohave
cycles
(mostbiologicalnetworkscontainfeedbackcycles)

DAGscanmodeleffectsof
interventions

Wecouldusedirectedacyclicgraphical(DAG)models:

Sowhatkindofgraphicalmodelshouldweuseforthisdata?

Wecoulduseundirectedgraphical(UG)models:

UGsallowthemodeltohave
cycles

ButUGsdonotmodeleffectsof
interventions
(thereisnodifferencebetween‘seeing’and‘doing’)

DrawbacksofDirectedAcyclicandUndirectedModels

noitubirtnoCruOnoitavitoMstnemirepxEnoitatnemelpmIledoMlaitnetoPlanoitnevretnInoitcudortnIButDAGsdonotallowthemodeltohave
cycles
(mostbiologicalnetworkscontainfeedbackcycles)

DAGscanmodeleffectsof
interventions

Wecouldusedirectedacyclicgraphical(DAG)models:

Sowhatkindofgraphicalmodelshouldweuseforthisdata?

Wecoulduseundirectedgraphical(UG)models:

UGsallowthemodeltohave
cycles

ButUGsdonotmodeleffectsof
interventions
(thereisnodifferencebetween‘seeing’and‘doing’)

DrawbacksofDirectedAcyclicandUndirectedModels

sledoMGCDgnisuataDlanoitnevretnIetercsiDgniledoMyhpruM.KdnatdimhcS.M