神经网络结构复杂度及其功能表现:梯径理论与算法信息论_第1页
神经网络结构复杂度及其功能表现:梯径理论与算法信息论_第2页
神经网络结构复杂度及其功能表现:梯径理论与算法信息论_第3页
神经网络结构复杂度及其功能表现:梯径理论与算法信息论_第4页
神经网络结构复杂度及其功能表现:梯径理论与算法信息论_第5页
已阅读5页,还剩10页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

npjComplexity

|(2024)1:151

npj|complexityArticle

/10.1038/s44260-024-00015-x

Correlatingmeasuresofhierarchical

structuresinartificialneuralnetworkswiththeirperformance

Checkforupdates

ZhuoyingXu1

,6

,YingjunZhu1

,6,BinbinHong

2

,XinlinWu3

,4

,5

,JingwenZhang3

,4

,5

,MufengCai3

,4

,5

,DaZhou

1

区&YuLiu

3

,4

ThisstudyemploystherecentlydevelopedLadderpathapproach,withinthebroadercategoryof

AlgorithmicInformationTheory(AIT),whichcharacterizesthehierarchicalandnestedrelationshipsamongrepeatingsubstructures,toexplorethestructure-functionrelationshipinneuralnetworks,

multilayerperceptrons(MLP),inparticular.Themetricorder-rateη,derivedfromtheapproach,isameasureofstructuralorderliness:whenηisinthemiddlerange(around0.5),thestructureexhibitstherichesthierarchicalrelationships,correspondingtothehighestcomplexity.Wehypothesizethatthehigheststructuralcomplexitycorrelateswithoptimalfunctionality.Ourexperimentssupportthis

hypothesisinseveralways:networkswithηvaluesinthemiddlerangeshowsuperiorperformance,andthetrainingprocessestendtonaturallyadjustηtowardsthisrange;additionally,startingneuralnetworkswithηvaluesinthismiddlerangeappearstoboostperformance.Intriguingly,thesefindings

alignwithobservationsinotherdistinctsystems,includingchemicalmoleculesandproteinsequences,hintingatahiddenregularityencapsulatedbythistheoreticalframework.

Itiswidelyrecognizedthattheperformanceofneuralnetworksdependsontheirarchitecture

1

.Ononehand,thedesignofthesestructuresisinspiredbybiologicalneuralnetworks.Forinstance,convolutionalnetworksdrawfromtheconceptofvisualreceptivefields

2

,3

.Ontheotherhand,theintroductionofnewnetworkarchitecturesoftenleadstosignificantperformanceleaps,suchastheTransformer,whichincorporatesaspectsofattentionmechanisms

4

.Whileitisknownthatthearchitectureofartificialneuralnetworksisakeydeterminantoftheirfunctionalityandperformance,inpractice,thedevelopmentofneuralnetworksismovingtowardsincreas-inglylargerscales,morecomplexstructures,andgreaternumbersoflayers.Forexample,thescaleofnetworks,intermsofthenumberofparameters,hasgrownfromthethousandsinLeNettothemillionsinAlexNet,andthentothebillionsinGPT;whilethearchitecturehasevolvedfromLeNettoAlexNet,VGG,GoogLeNet,variousAutoencoders,andmorerecentlytoTransformer

5

-9

.

Whilelarge-scalenetworksdemonstrateformidablecapabilitiesinpracticalapplications,theincreasedcomplexityofneuralnetworksimpliesaneedformoretrainingdata,longertrainingtimes,higherexpenses,andagrowingconcernforenergyconsumption

10

.Ifitwerepossibletoestimatetheappropriateneuralnetworkstructureandits

complexityforspecifictasksinadvance,substantialcomputationalresourcescouldbesaved.Additionally,insituationswherecompu-tationalresourcesarelimited,suchasinsmalldevicesorspaceequipment,itmightbenecessarytousethesmallestpossibleneuralnetworks

11

.Therefore,anunderstandingofthegenerallawsgov-erningtherelationshipbetweenthestructureandperformanceofneuralnetworksalsoholdssignificantpracticalvalue.

Researchersareactivelyinvestigatingtherelationshipbetweenneuralnetworkstructuresandtheirperformancefromvariousperspectives.Byrepresentingneuralnetworksasgraphs,onecanemployexistinggraphmetricsordefinenewonestoelucidatethelinkbetweennetworkperfor-manceandthesemetricsacrossdifferentapplications

12

,13

.Someresearchdirectlyfocusesontheparticularstructureofnetworks

14

-16

,whileotherstudiesintroducenovelneuralnetworkstructurestoexaminetheirimpactonnetworkperformanceenhancement

17

.Thesestudieshaveraisednumerousfundamentalquestionsthatareurgentlyinneedofbeinganswered.Questionssuchaswhetherthereisasystematicrelationshipbetweenneuralnetworkstructuresandtheirperformance,orwhetherhigh-performingneuralnetworkspossessdistinctstructuralcharacteristics,arecentraltothisresearcharea.

1SchoolofMathematicalSciences,XiamenUniversity,Xiamen,China.2DepartmentofPhysics,FacultyofArtsandSciences,BeijingNormalUniversity,

Zhuhai,China.3DepartmentofSystemsScience,FacultyofArtsandSciences,BeijingNormalUniversity,Zhuhai,China.4InternationalAcademicCenterof

ComplexSystems,BeijingNormalUniversity,Zhuhai,China.5SchoolofSystemsScience,BeijingNormalUniversity,Beijing,China.6Theseauthorscontributedequally:ZhuoyingXu,YingjunZhu.e-mail:

zhouda@

;

yu.ernest.liu@

/10.1038/s44260-024-00015-x

Article

Hierarchyisanimportantfeatureamongvariousstructuralchar-acteristics.Intheresearchintobothartificialandbiologicalneuralnetworks,theimportanceandthepotentialuniversalityofhierarchicalstructureshavebeendiscovered(seethetwoparagraphsattheendofthissection).However,quantitativelycharacterizingthehierarchicalstructureofneuralnetworksisnotatrivialtask.TheLadderpathapproach,arecentlydevelopedmathe-maticaltoolwithinthebroadercategoryofAlgorithmicInformationTheory(AIT),offersarigorousanalyticalapproachforexaminingthestructuralinformationofatargetsystem

18

,19

.

Thisapproachpinpointsrepeatingsubstructures(or“modules”)withinthetargetsystem,whichthemselvesmayconsistofevensmallerrepeatingunits,progressivelybreakingdowntothemostbasicbuildingblocks.Theserepeatingsubstructurescanthenbereorganizedintoahier-archical,modular-nested,tree-likestructure.Oneextremescenarioisasystemwithaflattenedhierarchy,essentiallydevoidofrepeatingstructures,andcanbelikenedtocompletelydisorderedstructuresorentirelyrandomgeneticsequences.Theotherextremeinvolvessimplerepetitionofsmallsubstructures,likeadoublingsequence(2becomes4,4becomes8,8becomes16,andsoforth),whichiscomparabletocrystals.Systemsdeviatingfromtheseextremesandpositionedinthemiddlearethosewitharichhierarchicalstructure.Wehypothesizethatthedegreeofrichnessinthishierarchicalstructureispositivelycorrelatedwiththeperformanceofneuralnetworks.

TheLadderpathapproachhasrecentlybeensuccessfullyappliedinlivingandchemicalsystems,suchasanalyzingtheevolutionofproteinsequences

19

;asimilarmethodhasalsobeenusedtoinvestigatetheoriginsoflifethroughtheanalysisofmolecularstructure

20

.Theapplicabilityofthisapproachintheseareasislinkedtothetightstructure-functionrelationshipprevalentinevolutionarysystems(e.g.,thestructureofmoleculesdictatestheirphysicochemicalproperties,andthestructureofdrugmoleculesdeterminestheirbindingwithreceptorproteins).Similarly,thisstructure-functionrelationshipislikelytoexistinanotherkeyevolutionarysystem:intelligence.ThispaperemploystheLadderpathapproachtoanalyzethestructure-functionrelationshipinartificialneuralnetworks.Weacknowl-edgethatourworkiscurrentlylimitedtomultilayerperceptrons(MLP)andnetworksofconstrainedsizesduetothetechnicalchallengesofladderpathcalculations,aswehaveshownthatthesecalculationsareanNP-hardproblem

18

.However,ourprimarymotivationistodemonstratetheeffec-tivenessandfeasibilityofthisAITapproachinstudyingthehierarchicalandnestedstructuresofartificialneuralnetworks.Moreimportantly,coupledwiththerecentworkonproteinsequencesasdemonstratedin

19

,thistypeof

structure-functionrelationshipappearstomanifestinseveralseemingly

unrelatedsystems,suggestingacertainuniversalitythatmeritsattention.

Beforemovingforward,letusreviewthepreviousliteratureonhier-archicalstructures.Thefirstpartisabouthierarchicalstructuresinartificialneuralnetworks.Forcertainspecificnetworks,suchasreservoircomputing,whichisanincreasinglyprominentneuralnetworkframeworkforstudyingdynamicalsystems

21

,researchhasshownthattheircomputationalcapacityismaximizedwhentheyareinacriticalstate

22

(althoughsomescholarsbelievethatthismaximizationofcomputationalcapacityisconditionalandonlytrueunderspecificcircumstances

23

).In2021,Wangetal.discoveredthatinwell-trainedreservoirnetworks,thenodessynchronizeinclusters,andthesizedistributionofthesynchronizationclustersfollowsapowerlaw

15

.Itisimportanttonotethatapower-lawdistributionisoftenakeycharacteristicofacriticalstate,andthisdistributionistypicallyamanifes-tationofarichlyhierarchicalstructure

24

.In2020,Leskovecetal.proposedamethodtorepresentneuralnetworksasgraphs,termedrelationalgraphs,whichcharacterizetheinter-layerconnectionsofneuralnetworks

13

.Theydiscoveredthatincaseswhereaneuralnetworkperformswell,theaveragepathlengthandclusteringcoefficientofitscorrespondingrelationalgraphtendtoalwaysfallwithinacertainrange(itisnotablethatthesetwoindicesintuitivelyrepresentcharacteristicsthataretypicallyoppositeinnature).Furthermore,thetrainingprocesstendstoshiftthesetwostructuralmetricstowardsthisrange.In2018,Yingetal.introducedadifferentiablegraphpoolingmethod,designedtointegratecomplexhierarchicalrepresentations

ofgraphs,andfurtherconnectdeepergraphneuralnetworkmodelstotheserepresentations

16

.Thismethodhassignificantlyenhancedaccuracyingraphclassificationbenchmarktaskscomparedtootherpoolingmethods,underscoringthecriticalroleofutilizinghierarchicalinformationforneuralnetworkperformance.In2016,Bengioetal.proposedthreecomplexitymetricsforthearchitectureofrecurrentneuralnetworksfromagraphperspective:recurrentdepth,feedforwarddepth,andrecurrentskipcoeffi-cient.Experimentshaverevealedthatincreasingrecurrentandfeedforwarddepthcanenhancenetworkperformance,andaugmentingtherecurrentskipcoefficientcanimproveperformanceintaskswithlong-termdependencies

12

.

Then,thisparagraphcovershierarchicalstructuresinbiologicalneuralnetworks.Inrealbiologicalneuralnetworks,similarstudieshavealsoshownapositivecorrelationbetweenhierarchicalstructureandgoodperformance.Baumetal.analyzedsamplesfromyouthsaged8–22inthePhiladelphiaNeurodevelopmentalCohorttostudytheevolutionofstructuralbrainnetworksovertime

25

.Theyobservedthatasindividualsage,thenetworkstructureundergoesamodularseparationprocess,whereconnectionswithinmodulesstrengthenandinter-moduleconnectionsweaken.Fur-thermore,thisseparationappearstobebeneficialforthedevelopmentofbrainfunctionsinyouths.Vidaurreetal.,usingwhole-brainresting-statefMRIdata,employedmethodssuchashiddenMarkovchainsandhier-archicalclusteringtoanalyzetheorganizationaldynamicsofbrainnetworksovertime

26

.Theydiscoveredadistincthierarchicalstructurewithinbraindynamics,whichshowssignificantcorrelationswithindividualbehaviorandgeneticpredispositions.Theirresearchunderscoresacloseconnectionbetweenthehierarchicalstructureandfunctioninactualbrainnetworks.In2021,Zhouetal.studiedtherelationshipbetweennetworkstructureanditsdynamicpropertiesfromdifferentperspectives.Theydiscoveredthatmodularnetworktopologiescansignificantlyreducebothoperationalandconnectivitycosts,thusachievingajointoptimizationofefficiency

27

.Alsoin2021,Luoreviewedcommoncircuitmotifsandarchitecturalplans,exploringhowthesecircuitarchitecturesassembletoachievevariousfunctions

28

.Thesecircuitmotifscanbeconsideredas“words”,whichcombineintocircuitarchitecturesthatmightoperateatthelevelof“sen-tences”.Thisworkemphasizestheimportanceofhierarchicalstructures:onlyaftergainingabetterunderstandingofthenestedmodularhierarchicalstructuresamongtheseneuronalconnectionscanwebegintounderstandthe“paragraphs”(e.g.,brainregions)andeventuallythe“article”,whichmayinspiremoreimportantadvancesinartificialintelligence.

Methods

RecaptheLadderpathapproach

TheLadderpathapproach,detailedin

18

,19

,offersaquantitativeapproachtoanalyzestructuralinformationinsystems,rangingfromsequences,mole-culestoimages.Ititerativelyidentifiesrepeatingsubstructures,calledlad-derons,whichareessentiallyreusedmodules.Thesesubstructuresmaybecomposedofsmallerrepeatingunits,cascadingdowntothesystem’smostbasicbuildingblocks.Theseladderonscollectivelyformahierarchical,nested,tree-likestructureknownasaladdergraph.

UsingthesequenceABCDBCDBCDCDACAC(denotedasX)asanexampletoillustrate,theladdergraphiscomputedasshowninFig.

1

,demonstratinghowtoconstructthetargetsequenceXfromthefourbasicbuildingblocksA,B,C,andDinthemostefficientmanner.First,combineCandDtoconstructCD,andAandCtoconstructAC,whichtakes2steps;thencombiningBwithCDtoconstructBCDisanotherstep.Next,con-structthetargetsequenceXusingalreadygeneratedsubstructures(or“modules”)andbasicbuildingblocks.ThisinvolvescombiningA,BCD,BCD,BCD,CD,AC,andAC,thustaking6steps;intotal,thisis9stepssofar;thefinalstepinvolvesoutputtingX.Therefore,constructingXinthemostefficientmannertakes10stepsintotal.

Xhas16letters,butonly10stepswereneededduetothereuseofsomerepeatingsubstructures:forexample,thepreviouslyconstructedCDisreusedinconstructingBCD;whileBCDappearsthreetimes,thesubsequentinstancesofBCDcandirectlyusetheinitiallyconstructedBCD,saving

npjComplexity

|(2024)1:152

/10.1038/s44260-024-00015-x

Article

Fig.1|Laddergraphsofdifferentsystems.aLaddergraphofashortsequence,forillustrativepurposes.Thegrayhexagonsrepresentthebasicbuildingblocks,andthegraysquaresrepresenttheladderons.Inb–d,basicbuildingblocksareomitted,andladderonsarerepresentedasgrayellipses.bLaddergraphofacompletelyrandomanddisorderedsequence,withη=0.04,BBDCDDCAACABACCDADA-

BABDDD...,whichshowsminimalhierarchicalrelationshipsamongrepeating

substructures.cLaddergraphofasequencecomposedentirelyofA's(themost

orderedsequence),withη=1.TheladderonsareAA,AAAA,AAAAAAAA,etc.dLaddergraphofthesequenceABABABDABABBAABAABCACABABDA...,

whichhasη=0.53.Thissequencedisplaystherichesthierarchicalstructure.Thesequencesshownin(b-d)areeachcomposedofthebasicbuildingblocksA,B,C,andD,andhavealengthof300,representingthreetypicalcategoriesofsequences.

manysteps.ThisprincipleofreuseisafundamentalconceptinAITandtheLadderpathapproach.

Notethatinourexample,somestepscanbeinterchanged,suchaswhethertoconstructCDorACfirst,whichdoesnotmatter.However,theorderbetweenCDandBCDmustnotbereversed,becauseBCDisbasedonCD.Therefore,theladdergraphalsocorrespondstoapartiallyorderedmultiset,whichcanbedenotedas

{B;D;A(2);C(2)/AC;CD/BCD(2)}(1)

Stepswithinthesamelevel(i.e.,separatedby“/”)canbeinterchanged,butnotacrosslevels.Thisiswhythesequenceexhibitsahierarchicalstructure.

Now,wecanintroduceseveralimportantnotionswithintheLadder-pathapproach.Intheexampleabove,the“10steps”aredefinedastheladderpath-indexofX,whileanotherquantity,theorder-indexω,isdefinedasthelengthofXminusitsladderpath-index,whichisω(X)=16-10=6.Mathematically,wehaveshowninref.

18

thatωalwaysequalsthesumofthe“reducedlengths”lofeachladderon(wherethereducedlengthisdefinedasthelengthofeachladderonminusone).Inthiscase,ω(X)=lAC+lCD+2×lBCDwherelACisthelengthofACminus1,namely(2−1),lCD=2−1,andlBCD=3−1;Themultiplierof2forlBCDisbecauseitsmultiplicityinthepartiallyorderedmultiset,asshowninEq.(

1

),is2,indicatingthatitwasreusedtwice.Thus,theorder-indexωcountsthesizesofallrepeatingsubstructuresinasystem,therebyessentiallychar-acterizinghoworderedasystemisinanabsolutemeasure.Foramoredetailedtheoreticalexplanationandmathematicalderivation,pleaserefertoref.

18

.

Thisrecapmaybeconsideredlengthy,whichcouldseemover-whelming,andsomenotionsmightappearunnecessaryatfirstglance.Infact,thisisbecausetheLadderpathapproachwithinAITwasnotspecificallydevelopedtocharacterizeneuralnetworksbutwasdevelopedfromamoregeneralperspective,whichmaymakeitseemverbose.However,ithasbeensuccessfullyappliedinseeminglyunrelatedfieldssuchasproteinsequences

19

,andoursubsequentfindingsalsodemonstratethatthisapproachperformswellinartificialneuralnetworks.Thisindicatesthattheapproachandtherelationshipsbetweenthehierarchicalstructureand

performanceitdescribesareuniversaltoacertainextent,makingitworthwhile.

Ladderpathcharacterizeshierarchicalstructures

Aftertherecap,weproceedbytakingmorecomplexsequencesasexamples.Wewilldemonstratetheladdergraphsofthreetypicalcategoriesofsequences:minimalrepetition,akintocompletelyrandom,disorderedsystems(Fig.

1

b);simplerepetition,similartoacrystallinestructure,wherethepatternprogressesfrom2to4,4to8,8to16,andsoon(Fig.

1

c);andtheonesthatliebetweenthesetwoextremes,showcasingtherichesthierarchicalstructure(Fig.

1

d).

Tobettercharacterizethehierarchicalstructure,wehave,inref.

19

,definedarelativemeasureontopoftheabsolutemeasureω,calledtheorder-rateη.Tobeginwith,letusfirstexaminethedistributionofωofvarioussequencesversustheirlengthsS,asshowninFig.

2

.ForagivenlengthS,wecanobservethatωhasbothamaximumandaminimumvalue:themax-imum,denotedasωmax(S),correspondstosequencesthatarecompletelyidentical;theminimum,denotedasω0(S),correspondstopurelyrandomsequences.Theminimumvaluearisesbecause,insequenceswithafinitenumberofbases(herebeingA,B,C,andD),repeatingsubstructureswillinevitablyappearasthelengthincreases,meaningeventhepurelyrandomsequencewillnothaveanωofzero.Therefore,weneedtonormalizeit,leadingtothedefinitionoftherelativemeasureorder-rateη(x)forsequencex:

η(x):=ω(x)-ω0(S)(2)

ωmax(S)-ω0(S)

whereSisthelengthofthesequencex.Anηof0meanscompletedisorder(Fig.

1

b),1impliesfullorder(Fig.

1

c),andaround0.5indicatesarichlystructuredhierarchy(Fig.

1

d).

Infact,intheLadderpathapproach,ηisrelatedtothecomplexityofasystem.Asystemisnotconsideredcomplexifentirelyrandom(η≈0)orordered(η≈1);complexityemergesonlyintheintermediatestate(η≈0.5).ThisdistinguishesLadderpathfromsimilarconceptssuchasKolmogorov

npjComplexity

|(2024)1:153

/10.1038/s44260-024-00015-x

Article

npjComplexity

|(2024)1:154

complexity,additionchain,assemblytheory,andthe“adjacentpossible”

29

–31

.

Inthecontextofneuralnetworks,wecanapplytheLadderpathapproachtosystematicallyorganizethenetwork’srepeatingsub-structuresinanestedhierarchicalmanner.AsillustratedinFig.

3

a,b,thepartshighlightedbyred,yellow,andgreenlinesrepresentthesesubstructures(withthesamecolorindicatingidentical,reusedmodules).Fromthis,wecanestablishahierarchicalrelationship(andillustratetheladdergraph):theredsubstructureisencompassedwithinboththeyellowandgreenones.Nevertheless,directlycalcu-latingtheladdergraphisquitechallenging(andthisproblemisinherentlyNP-hard

18

).Hence,wefirsttransformthenetworkintoa

Fig.2|Distributionoftheorder-indexωforsequencesofvaryinglengths,illus-tratingthecalculationoftheorder-rateη.

setofsequencesandthencomputeit(themethodforthistrans-formationwillbedetailedinsection“Sequencerepresentationofaneuralnetwork”,andtherearealreadyalgorithmsdevelopedforcomputingladdergraphsofsequencesatthescaleof10,000inlength

19

).Finally,wehypothesizethattheneuralnetwork’sabilitytoextractandintegrateinformationisatitspeak(achievingitsbestperformance)whenitshierarchicalstructureistherichest(i.e.,whenmodulereuseandtinkeringaremostpronounced),correspondingtotheorder-rateηaround0.5.

Experimentsetup

Toinvestigatetheconnectionbetweenthestructureofaneuralnetworkanditsfunctionality,wechosearathersimpletask:recognizingwhetherathree-digitnumberisoddoreven.ThiswasaddressedusingaMLP.Theinputconsistsofthreeneurons,representingthehundreds,tens,andonesplace,respectively,whiletheoutputhastwoneuronsindicatingoddoreven.Thehiddenlayersvary,either1,2,3,or4layers,eachcomprisingadifferentnumberofneurons.Tosimplifytheanalysis,welimitedeachMLPtoamaximumof200edges.Withthisconstraint,anMLPwithonehiddenlayercouldhaveonly40distinctvariations:Giventheinputlayerhas3neuronsandtheoutputlayerhas2neurons,ifthehiddenlayerhas40neurons(wecandenotethisMLP[3,40,2]),thereareatotalof3×40+40×2=200edges;theMLP[3,41,2]wouldhave205edges.ForanMLPwithtwo,three,andfourhiddenlayers,weconstructed200distinctarchitecturesforeachcategory(suchas[3,7,17,2],[3,8,3,5,2],and[3,3,5,14,2]whichhave2,3,and3hiddenlayersrespectively),contributingtothetotalof640differentarchitectures.

Thenetworkconnectionweightswererandomlyinitialized,andall640architecturesunderwentanidenticaltrainingperiod,eachfor2000epochs.Duringthetraining,wemonitoredthechangesinperformance(measuredbyaccuracy)andthenetwork’sorder-rateη(calculatedbasedontheLad-derpathapproach),toexploretherelationshipbetweentheseaspects.Sec-tion“Results”presentstheevolutionaryandstatisticalrelationshipsbetweenstructureandfunction.

Fig.3|ThediagramillustratinghowtoemploytheLadderpathapproachtoanalyzeaneuralnetwork,andreorganizetherepeatingsubstructuresintoaladdergraph.aAschematicdiagramofanMLP.bAschematicdiagramoftheladdergraphofthisMLP.cRepresentinganMLPasasetofsequences.

/10.1038/s44260-024-00015-x

Article

Sequencerepresentationofaneuralnetwork

ToutilizethealgorithmbasedontheLadderpathapproach

19

forstudyingthestructure-functionrelationshipinneuralnetworks,weneedtouseasequenceorasetofsequencestorepresentaneuralnetwork’sstructure.Byenvisioninganeuralnetworkasasignal(i.e.,theinput)propagationpro-blemoraninformationflowproblemwithinanetwork,thecollectionofallpathsthatthesesignalstraversecanrepresentthenetworkstructure.Sincetheconnectionweightsbetweenneuronsarerealnumbers,wefirstneedtocoarse-grainthem,usingdifferentsymbolstorepresentweightswithinthesamerange.

Thehigherthedegreeofcoarse-graining,themoreconducivetheextractedinformationisforsequenceanalysis;thelowerthedegreeofcoarse-graining,themoreinformationretained.However,thisalsoresultsinmoreunnecessarydetails,whichcanmakesubsequentanalysismoredif-ficult.Therefore,weconductedexperimentswithvariousdegreesofcoarse-grainingtofindabalancethatisaslargeaspossiblewithoutsignificantlydiminishingfunctionalperformance.Fortheparticularsystemsweselected,ournumericalexperimentsshowedthatwhenthecoarse-grainingintervalis

Order-rateη

Fig.4|Thedistributionoftheorder-rateηvs.theaccuracyintheodd-even

recognitiontaskperformedbytheneuralnetworks.Thestatisticsincludealloftheaforementioned640differentarchitectures.

settoamaximumof0.1,ithasaminimalimpactontheneuralnetwork’sperformance(seeSupplementaryNote1fordetailedinformation).

Subsequently,weobtaintheappropriatelycoarse-grainedgraph,andwecanconvertthegraphintoasetofsequences.Neuronsarethenodesofthisgraph,andnodeswithinthesamelayerareassignedthesamesymbolduetotheirsharedactivationfunction.Connectionsbetweenneuronsaretheedgesofthegraph,andedgeswithuniqueweights(discretizedaspre-viouslymentioned)areconsideredtocarrydistinctinformation,andarethusrepresentedusingdifferentsymbols.Forexample,thepathhighlightedingrayinFig.

3

canbedenotedasAzBxC.Throughthismethod,wecandetaileverypathaninputsignaltraverses,andaggregatethesepathsintoasetofsequencestodepictthegraph.WecanthenemploytheLadderpathapproachtoexaminethesequences,andtherebyinvestigatethechar-acteristicsoftheneuralnetwork’sstructure.

Results

Bestperformancewhenthehierarchicalstru

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论