简介:Anewmethodwasdescribedforusingarecurrentneuralnetworkwithbiasunitstopredictcontactmapsinproteins.Themaininputstotheneuralnetworkincluderesiduespairwise,residueclassificationaccordingtohydrophobicity,polar,acidic,basicandsecondarystructureinformationandresidueseparationbetweentworesidues.Inourwork,adatasetwasusedwhichwascomposedof53globulinproteinsofknown3Dstructure.Anaveragepredictiveaccuracyof0.29wasobtained.Ourresultsdemonstratetheviabilityoftheapproachforpredictingcontactmaps.
简介:TheBIONISnetworkwassetupinspring2002byProf.GeorgeJeronimidis,Prof.JulianVincentandPhilSheppard.Membershipisnowover300,withmembersfromacademiaandindustryinmorethan40countries.ThemissionofthenetworkistopromotetheapplicationofBiomimeticsinproductsandservicesanditsuseineducationandtraining.ItiscurrentlysupportedbySwedishBiomimetics3000?andhostedbytheUniversityofReading.
简介:Simulatingbiologicalolfactoryneuralsystem,KⅢnetwork,whichisahigh-dimensionalchaoticneuralnetwork,isdesignedinthispaper.Differentfromconventionalartificialneuralnetwork,theKⅢnetworkworksinitschaotictrajectory.ItcansimulatenotonlytheoutputEEGwaveformobservedinelectrophysiologicalexperiments,butalsothebiologicalintelligenceforpatternclassification.Thesimulationanalysisandapplicationtotherecognitionofhandwritingnmeralsarepresentedhere.TheclassificationperformanceoftheKⅢnetworkatdifferentnoiselevelswasalsoinvestigated.
简介:Anactivestereovisionsystembasedonamodelofneuralpathwaysofhumanbinocularmotorsystemisproposed.Withthismodel,itisguaranteedthatthetwocamerasoftheactivestereovisionsystemcankeeptheirlinesofsightfixedonthesametargetobjectduringsmoothpursuit.Thisfeatureisveryimportantforactivestereovisionsystems,sincenotonly3Drecon-structionneedsthetwocamerashaveanoverlappingfieldofvision,butalsoitcanfacilitatethe3Dreconstructionalgorithm.Toevaluatetheeffectivenessoftheproposedmethod,somesoftwaresimulationsaredonetodemonstratethesametargettrackingcharacteristicinavirtualenvironmentapttomistrackingeasily.Here,mistrackingmeanstwoeyestracktwodifferentobjectsseparately.Thentheproposedmethodisimplementedinouractivestereovisionsystemtoperformrealtrackingtaskinalaboratoryscenewhereseveralpersonswalkself-determining.Beforetheproposedmodelisimplementedinthesystem,mis-trackingoccurredfrequently.Afteritisenabled,mistrackingneveroccurred.Theresultshowsthatthevisionsystembasedonneuralpathwaysofhumanbinocularmotorsystemcanreliablyavoidmistracking.
简介:Inthepost-genomicbiologyera,thereconstructionofgeneregulatorynetworksfrommicroarraygeneexpressiondataisveryimportanttounderstandtheunderlyingbiologicalsystem,andithasbeenachallengingtaskinbioinformatics.TheBayesiannetworkmodelhasbeenusedinreconstructingthegeneregulatorynetworkforitsadvantages,buthowtodeterminethenetworkstructureandparametersisstillimportanttobeexplored.Thispaperproposesatwo-stagestructurelearningalgorithmwhichintegratesimmuneevolutionalgorithmtobuildaBayesiannetwork.Thenewalgorithmisevaluatedwiththeuseofbothsimulatedandyeastcellcycledata.Theexperimentalresultsindicatethattheproposedalgorithmcanfindmanyoftheknownrealregulatoryrelationshipsfromliteratureandpredicttheothersunknownwithhighvalidityandaccuracy.