000912163 001__ 912163
000912163 005__ 20240313103120.0
000912163 0247_ $$2doi$$a10.1103/PhysRevResearch.4.043143
000912163 0247_ $$2Handle$$a2128/32946
000912163 0247_ $$2WOS$$aWOS:000933947400011
000912163 037__ $$aFZJ-2022-05381
000912163 082__ $$a530
000912163 1001_ $$0P:(DE-Juel1)180150$$aFischer, Kirsten$$b0$$eCorresponding author
000912163 245__ $$aDecomposing neural networks as mappings of correlation functions
000912163 260__ $$aCollege Park, MD$$bAPS$$c2022
000912163 3367_ $$2DRIVER$$aarticle
000912163 3367_ $$2DataCite$$aOutput Types/Journal article
000912163 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1670307479_6446
000912163 3367_ $$2BibTeX$$aARTICLE
000912163 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000912163 3367_ $$00$$2EndNote$$aJournal Article
000912163 520__ $$aUnderstanding the functional principles of information processing in deep neural networks continues to be a challenge, in particular for networks with trained and thus nonrandom weights. To address this issue, we study the mapping between probability distributions implemented by a deep feed-forward network. We characterize this mapping as an iterated transformation of distributions, where the nonlinearity in each layer transfers information between different orders of correlation functions. This allows us to identify essential statistics in the data, as well as different information representations that can be used by neural networks. Applied to an XOR task and to MNIST, we show that correlations up to second order predominantly capture the information processing in the internal layers, while the input layer also extracts higher-order correlations from the data. This analysis provides a quantitative and explainable perspective on classification.
000912163 536__ $$0G:(DE-HGF)POF4-5232$$a5232 - Computational Principles (POF4-523)$$cPOF4-523$$fPOF IV$$x0
000912163 536__ $$0G:(DE-HGF)POF4-5234$$a5234 - Emerging NC Architectures (POF4-523)$$cPOF4-523$$fPOF IV$$x1
000912163 536__ $$0G:(DE-Juel-1)BMBF-01IS19077A$$aRenormalizedFlows - Transparent Deep Learning with Renormalized Flows (BMBF-01IS19077A)$$cBMBF-01IS19077A$$x2
000912163 536__ $$0G:(DE-Juel1)HGF-SMHB-2014-2018$$aMSNN - Theory of multi-scale neuronal networks (HGF-SMHB-2014-2018)$$cHGF-SMHB-2014-2018$$fMSNN$$x3
000912163 536__ $$0G:(DE-HGF)SO-092$$aACA - Advanced Computing Architectures (SO-092)$$cSO-092$$x4
000912163 536__ $$0G:(DE-82)EXS-SF-neuroIC002$$aneuroIC002 - Recurrence and stochasticity for neuro-inspired computation (EXS-SF-neuroIC002)$$cEXS-SF-neuroIC002$$x5
000912163 588__ $$aDataset connected to CrossRef, Journals: juser.fz-juelich.de
000912163 7001_ $$0P:(DE-Juel1)178936$$aRené, Alexandre$$b1$$ufzj
000912163 7001_ $$0P:(DE-Juel1)171384$$aKeup, Christian$$b2
000912163 7001_ $$0P:(DE-Juel1)174497$$aLayer, Moritz$$b3
000912163 7001_ $$0P:(DE-Juel1)156459$$aDahmen, David$$b4
000912163 7001_ $$0P:(DE-Juel1)144806$$aHelias, Moritz$$b5
000912163 773__ $$0PERI:(DE-600)3004165-X$$a10.1103/PhysRevResearch.4.043143$$gVol. 4, no. 4, p. 043143$$n4$$p043143$$tPhysical review research$$v4$$x2643-1564$$y2022
000912163 8564_ $$uhttps://juser.fz-juelich.de/record/912163/files/Fischer_2022_prr.pdf$$yOpenAccess
000912163 8564_ $$uhttps://juser.fz-juelich.de/record/912163/files/PhysRevResearch.4.043143.pdf$$yOpenAccess
000912163 8767_ $$8INV/22/OCT/009522$$92022-10-10$$a1200185892$$d2022-11-15$$eAPC$$jZahlung erfolgt$$zUSD 2625,- ; FZJ-2022-04464
000912163 909CO $$ooai:juser.fz-juelich.de:912163$$pdnbdelivery$$popenCost$$pVDB$$pdriver$$pOpenAPC$$popen_access$$popenaire
000912163 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)180150$$aForschungszentrum Jülich$$b0$$kFZJ
000912163 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)178936$$aForschungszentrum Jülich$$b1$$kFZJ
000912163 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)171384$$aForschungszentrum Jülich$$b2$$kFZJ
000912163 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)174497$$aForschungszentrum Jülich$$b3$$kFZJ
000912163 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)156459$$aForschungszentrum Jülich$$b4$$kFZJ
000912163 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)144806$$aForschungszentrum Jülich$$b5$$kFZJ
000912163 9131_ $$0G:(DE-HGF)POF4-523$$1G:(DE-HGF)POF4-520$$2G:(DE-HGF)POF4-500$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-5232$$aDE-HGF$$bKey Technologies$$lNatural, Artificial and Cognitive Information Processing$$vNeuromorphic Computing and Network Dynamics$$x0
000912163 9131_ $$0G:(DE-HGF)POF4-523$$1G:(DE-HGF)POF4-520$$2G:(DE-HGF)POF4-500$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-5234$$aDE-HGF$$bKey Technologies$$lNatural, Artificial and Cognitive Information Processing$$vNeuromorphic Computing and Network Dynamics$$x1
000912163 9141_ $$y2022
000912163 915pc $$0PC:(DE-HGF)0001$$2APC$$aLocal Funding
000912163 915pc $$0PC:(DE-HGF)0002$$2APC$$aDFG OA Publikationskosten
000912163 915__ $$0StatID:(DE-HGF)0510$$2StatID$$aOpenAccess
000912163 915__ $$0LIC:(DE-HGF)CCBY4$$2HGFVOC$$aCreative Commons Attribution CC BY 4.0
000912163 915__ $$0StatID:(DE-HGF)0200$$2StatID$$aDBCoverage$$bSCOPUS$$d2022-11-29
000912163 915__ $$0StatID:(DE-HGF)0300$$2StatID$$aDBCoverage$$bMedline$$d2022-11-29
000912163 915__ $$0StatID:(DE-HGF)0501$$2StatID$$aDBCoverage$$bDOAJ Seal$$d2022-08-16T10:08:58Z
000912163 915__ $$0StatID:(DE-HGF)0500$$2StatID$$aDBCoverage$$bDOAJ$$d2022-08-16T10:08:58Z
000912163 915__ $$0StatID:(DE-HGF)0030$$2StatID$$aPeer Review$$bDOAJ : Blind peer review$$d2022-08-16T10:08:58Z
000912163 915__ $$0StatID:(DE-HGF)0199$$2StatID$$aDBCoverage$$bClarivate Analytics Master Journal List$$d2022-11-29
000912163 915__ $$0StatID:(DE-HGF)0112$$2StatID$$aWoS$$bEmerging Sources Citation Index$$d2022-11-29
000912163 915__ $$0StatID:(DE-HGF)0150$$2StatID$$aDBCoverage$$bWeb of Science Core Collection$$d2022-11-29
000912163 920__ $$lyes
000912163 9201_ $$0I:(DE-Juel1)INM-6-20090406$$kINM-6$$lComputational and Systems Neuroscience$$x0
000912163 9201_ $$0I:(DE-Juel1)IAS-6-20130828$$kIAS-6$$lTheoretical Neuroscience$$x1
000912163 9201_ $$0I:(DE-Juel1)INM-10-20170113$$kINM-10$$lJara-Institut Brain structure-function relationships$$x2
000912163 9801_ $$aFullTexts
000912163 980__ $$ajournal
000912163 980__ $$aVDB
000912163 980__ $$aUNRESTRICTED
000912163 980__ $$aI:(DE-Juel1)INM-6-20090406
000912163 980__ $$aI:(DE-Juel1)IAS-6-20130828
000912163 980__ $$aI:(DE-Juel1)INM-10-20170113
000912163 980__ $$aAPC
000912163 981__ $$aI:(DE-Juel1)IAS-6-20130828