CCS355 - NEURAL NETWORK AND DEEP LEARNING
QUESTION BANK
UNIT-I
PART-A
Q1.Define Neural Networks?
Q2.Define Neuromorphic Computing.
Q3.What are dentrites?
Q4.What is the function of dentrite?
Q5.What is axon?Give its function.
Q6. What is short-term memory in neural networks?
Q7. What is Long-term memory in neural networks?
Q8.What are learning laws in neural networks?
Q9.List the different learning laws in neural networks.
Q10. What are the requirements of learning laws?
Q11.What is learning rate parameter?
Q12.State the Hebbian learning law.
Q13.What are the features of neural networks?
Q14.What is stability-plasticity dilemma?
Q15.What are the characteristics of logistic functions?
Q16.Give the limitation of McCulloch Pitt’s neuron model.
Q17.Give the limitation of Rosen Blatt’s perceptron neuron model.
Q18.What is short-term memory in ANN?
Q19.What is Long-term memory in ANN?
Q20.Draw the structure of a biological neuron.
Q21.How is the net input calculated using matrix multiplication
method?
1
Q.How Adalinen euron model is different from McCulloch Pitt’s and RosenBlatt’s perceptron
model?
Q.ExemplifythemotivationforANN.
Q.Exemplifythesignificanceofsigmoidfunction.
Q.Forthenetworkshowninfigure,CalculatethenetinputtotheoutputneuronY.
Q.WhatislearninginANN?ListtheimportantlearningstrategiesinANN.
Q.Whysigmoidfunctionisalsocalledassquashingfunction?
Q.GivetheterminologyofArtificialNeuralNetworks.
Q.Drawthestructureofabiologicalneuronandalso,labelitsparts.
Q.Statefewactivationfunctionswhichareusedinsingleandmultilayernetworkstocalculatethe
output.
Q.WhatisthesignificanceofweightsusedANN.
Q.Inwhatareastheartificialneuralnetworksareused?
Q.Distinguish between Biologicalneural networkand artificialneural networks.
Q.Exemplifylinearly separableproblem.
Q.ShowthemappingandstructuralviewpointofANN.
Q.DiscussbrieflyMcCullochPitt’sartificialneuronmodel.Giveitslimitations.
Q.Givethemathematicalrepresentationof Rosen Blatt’s perceptron neuron model.
Q.GivethemathematicalrepresentationofWidrow’sAdalineneuronmodel.
Q.Identifyandillustratedifferentactivationfunctionsusedinneuralnetworks.
Q. Showthelinearizationof sigmoidfunction usingTaylor’s series.
Q.ListtheimportantlearninglawsinANN.DiscussbrieflyHebbianlearning.
Q.Graphically,sketchthedifferentactivationfunctionsusedinNN.
Q.GivethetaxonomyofdevelopmentsinArtificialNeuralNetworks.
PART-B
Q.Showthegraphicalrepresentationofsigmoidfunctionanddifferentiatethesigmoidfunctionandc
ommentontheresult.
Q.Showthatthesigmoidfunctionisrelatedtoothermathematicalfunctions.
Q.ObtaintheoutputoftheneuronYforthenetworkshowninfigureusingactivationfunctionasbi
narysigmoidalandBipolarsigmoidal.
Q.Distinguish betweenSupervisoryLearningandUnsupervisoryLearninginANN.
Q.HighlightbrieflytheHistoricaldevelopmentofArtificialNeuralNetworks.
Q.Discussbrieflythestructureandfunctionofabiologicalneuron.
Q.Listandexplaindifferentlearninglaws.
Q.RealizeNOTfunctionusingMcCulloch-Pittsneuronmodel.
Q.GeneratetheoutputoflogicANDfunctionbyMcCulloch-Pittsneuronmodel.
Q.GeneratetheoutoflogicORfunctionbyMcCulloch-Pittsneuronmodel.
Q.RealizetheEx-ORfunctionbyusingtheMcCulloch-Pittsneuronmodel.
UNIT2
PART-A
Q. What is Autoassociative Memory Network
Q.Whatisgeneralization?
Q.WhatisNetworkPruning?
Q.Whatismomentum?
Q.Howtodeterminethenumberhiddenneuronsinsinglehiddenlayerfeed-
forwardneuralnetwork?
Q. Listanytwoapplicationsof BackPropagationnetwork.
Q.MentionthedemeritsofBackpropagationNetwork.
Q.MentionthemeritsofBackpropagationNetwork.
Q.Sketchthe architectureof10/16/5multiplayerfeedforward neuralnetwork.
Q.NarratethedifferentapproachesandmethodsofimprovinggeneralizationinfeedforwardNeural
Networks.
Q. Feed-ward neural network for a given application uses 10 neurons in its input
layerand 5 neurons in its output layer. Determine the possible number of hidden
neuronsrequired in the hidden layer. Assume that the network require only one hidden
layer forthegivenapplication.
Q.For the givenapplication, thefeed-forwardneural networkof 10/12/6
isused.Determineitssize,widthanddepthofthenetwork.
Q.ListtheapplicationsofBackpropagationNetwork.
Q.ShowthelinearizationofSigmoidFunction.
Q.Whatisnetworkpruning?Mentiondifferentpruningtechniques.
Q.Howtodeterminethenumberhiddenneuronsinsinglehiddenlayerfeed-
forwardneuralnetwork?Explainwithanexample.
Q.Mention themerits and demeritsofBack propagationnetwork.
Q.Distinguishbetweenpatternsbasedlearningandbatchmodelearning.
Q.Give thesignificanceofmomentumintrainingfeed-
forwardneuralnetwork.
Q.GivetheflowchartfortheGeneralizedDeltaRule(GDR)procedure.
Q.Distinguishbetweenonlinelearningandoff-linelearning.
Q.Givethestep-by–stepprocedureofBack-PropagationAlgorithm
thatusesGDRprocedureforthreelayerfeed-forwardneuralnetwork.
PART-B
Q.Writetheflowchartoferrorback-propagationtrainingalgorithm.
Q.Explainbrieflytheoperationof10/16/5multiplayerfeedforwardneuralN/W
Q.Whatisgeneralization?Howtomeasure andevaluate generalization?
Q.Whatiscross-validation?Giveitssignificanceinfeedforwardn/wdesign.
Q.DiscussbrieflythepracticalconsiderationsinthedesignandimplementationofMultiLayerPercep
tronFeedForwardNeuralNetwork.
Q.WhatdoyouunderstandbyCascadeCorrelationArchitecture?Explainbriefly.
Q.DevelopaBackpropagationalgorithmforMultilayerFeedforwardneuralnetworkconsistingofon
einputlayer,onehiddenlayerandoutputlayerfromfirstprinciples.
Q.Discussbrieflythepracticalconsiderationsinimplementationoffeed-
forwardneuralnetworks.
Q.Listthefactorsthataffecttheperformanceofmultilayerfeed-forwardneuralnetwork.
UNIT3
PART-A
Q. What is Spiking Neural Networks?
Q.Whatisrecurrentneuralnetwork?
Q.WhatisHammingDistance?
Q.WhatisAssociativeMemory?
Q.WhatisContentAddressableMemory?
Q.NamethetwotypesofAssociativememory.
Q.NamethetwotypesofBidirectionalAssociativeMemory.
Q.WhatisPatternAssociationProblem?
Q.Whatissimulated-annealing?
Q.WhatisBoltzmannMachine?
Q.ListthemeritsofBoltzmann’sMachine.
Q.Mention thetwo differentarchitectures ofBoltzmann Machine.
Q.ListtheapplicationsofBoltzmann’sMachine
Q.WhatisHopfieldMemory?
Q.NamethetwoalgorithmsdevelopedforPatternAssociationnetworks.
Q.DeterminetheHammingdistancebetweenthetwovectorsgivenbelow.A=[0,
1,0,0,1,1,1]andB=[0,1,1,0,1,0,1].
Q.Atypicalpartitioningproblemconsistingof1000vectorsand20non-
emptysubsets.Determinethenumberofpossiblepartitions.
Q.Writetheexpressiontodeterminethe approximate possible numberofPartitions
inthecontextofclusteringproblem.
Q. Givethegeneralnetworksynthesisprocedure.
Q.ExplainbrieflyClusteringcomplexity
Q.GivetheweightstorageprescriptionforDiscreteBAMforbinaryinputvectorandbipolarinputvec
tor.
Q.What arethebasic parameters of recurrent network design?
Q.HowtodeterminethestoragecapacityofBAM?
Q.Listthemerits,demeritsandapplicationsofBoltzmann’sMachine.
Q.Considerthetwovectorsx=[1.-1,1-1]andy=[1,-1,1,-
1].ComputetheHammingdistancebetweenthem.
Q.DeterminetheHammingdistanceandaverageHammingDistancebetweenthegivenb
elowthetwovectorsA=[0,1,0,0,1,1,1]andB=[0,1,1,0,1,0,1].
Q.Showthe typicalarchitecture ofDiscreteHopfieldMemory.
Q.Whatisrecurrentneuralnetwork?Givetheclassificationofrecurrentn/w.
Q.DistinguishbetweenContinuousBAMandDiscreteBAM.
Q.DistinguishbetweenAutoassociativeMemoryandHeteroAssociativeMemory.
Q.DiscussbrieflytheworkingconceptsofBoltzmannMachine.
Q.WhatisHopfieldMemory?Explainbriefly.
Q.Whatiscontentaddressablememory(CAM)?Explainbriefly.
Q.StatetheOuter-ProductRule.
Q.Whatisbi-directionalmemory(BAM)?Explainbrieflywithitsarchitecture.
Q.ProvidethetrainingalgorithmforHopfieldMemory.
Q.Showthestep-by-
steptrainingalgorithmforbasicpatternassociationproblemusingHebbrule.
Q.AheteroassociativenetworkistrainedbyHebbouterproductruleforinputvectorS=[x1,x2,x3,x
4]tooutputrowvectorst=[t1,t2].Findtheweightmatrix:
S1=(1,1,0,0),S2=(1,1,1,0), S3=(0,0,1,1),S4=(0,1,0,0)
t1=(1,0),t2=(0.1),t3 =(1,0),t4=(1,0)
Q.Aheteroassociativenetworkisgiven.Findtheweightmatrixandtestthenetworkwiththetrain
inginputvectors.
S1=(1,1,0,0),S2=(0,1,0,0), S3=(0,0,1,1),S4=(0,0,1,0)
t1=(1,0),t2=(1,0),t3 =(0,1),t4=(0,1)
Q.Givethestep-by-steptrainingandtestingprocedureofDiscreteBAM
UNIT4
PART-A
Q. What is deep feedforward networks?
Q.Whatarecompetitiveneuralnetworks?
Q.WhatisSelf-Organization?
Q.ShowtheStructureofMAXNETneuralnetwork.
Q.ShowthestructureofKohonen’sSOFM.
Q.What arecompetitiveneural networks? Showitstypicalarchitecture.
Q.ProvidethesummaryofSOFMalgorithm.
Q.WritetheflowchartofKohenenSOFM.
Q.ProvidetheprocedureforC-meansalgorithm.
Q.WhatareSelf-organizingneuralnetworks?Explainbriefly.
Q. Discuss briefly the formal characterization clustering and
generalclusteringprocedures.
Q.ExplainbrieflywithaneatdiagramtypicalarchitectureofSOFM.
Q.WhatisMax-Net/MAXNET?Explainbriefly.
PART-B Q.Providet
he
trainingalg
orithmforMAXNET.
Q.Sketchthearchitectureofd=4MAXNETandexplainitsoperationbriefly.
Q.WritethestructureofKohenen’sSOFMnetworkandgiveitsalgorithmicsteps.
Q.AKohonenselforganizingmapisshownwithweightsinFiggivenbelow:
C1 C2 C3 C4 C5
U U
(a) UsingthesquareoftheEuclideandistancefindtheclusterunitCjthatisclosesttotheinput
vector(0.3,0.4)
(b) Usingalearningrateof0.3,findthenewweightsforunitCj
(c) FindnewweightsforCj-1andCj+1,iftheyareallowedtolearn.
Q.ConsideraKohonennetworkwithtwoclusterunitsandfiveinputunits.Theweighvectorsfort
heclusterunitsare
w1=[0.1,0.3,0.5,0.7,0.9]
and
w2=[0.9, 0.7, 0.5,0.3,0.1]
UsethesquareoftheEuclideandistancetofindthewinningclusterunitfortheinputpattern.
UNIT5
PART-A
Q. What is Recursive Neural Networks
Q.WhatisRBFNeuralNetwork?
Q.Whatis Time-DelayNeural Networks?
Q.Mentionthe twolayers inART1.
Q.Givetheadvantage ofAdaptiveResonanceTheorybased neuralnetworks.
Q.Mentionthe applicationsofTime-DelayNeuralNetworks.
Q.MentiontheapplicationsofAdaptive ResonanceTheorybasedneuralnetworks.
Q.Giveanyone applicationofRBFNeuralNetworkandTime-DelayNeuralNetwork.
Q.ListthemainfeaturesofRBFNN.
Q.SketchthearchitectureofatypicalRBFNetwork.
Q.SketchthearchitectureofatypicalTime-Delay–NeuralNetwork(TDNN).
Q.StateCover’stheoreminthecontextofRBFNeuralNetwork.
Q.ShowthestructureofatypicalRBFneuralnetwork.
Q.Whatisstability-plasticityDilemma?Explainbriefly.
Q.Show theArchitectureofART1.
PART-B
Q.Whatisadaptiveresonancetheory?Explainbriefly.
Q.DiscussbrieflythearchitectureofTime-DelayNeuralNetwork(TDNN).
Q.ListthetechniquesusedtoupdatetheweightsandcentresofRBFN.
Q.GivethetrainingprocedureforRBFNN.
Q.DrawthetypicalarchitectureofART-1andexplainitsoperation.
Q.DistinguishbetweenRadianBasisFunctionNeuralNetworkandMultilayerPerceptr
onFeed-ForwardNeuralNetwork.
Q.Providethe trainingalgorithmforthe RBFNNwithitsflowchart.
