公开(公告)号：EP3991102A1

公开(公告)日：2022-05-04

申请号：EP20734085.2

申请日：2020-06-26

申请人： Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V.

发明人： SAMEK, Wojciech ,  LAPUSCHKIN, Sebastian ,  WIEDEMANN, Simon ,  SEEGERER, Philipp ,  YEOM, Seul-Ki ,  MÜLLER, Klaus-Robert ,  WIEGAND, Thomas

IPC分类号： G06N3/08 ,  G06N3/04

公开(公告)号：EP3271863A1

公开(公告)日：2018-01-24

申请号：EP15711738.3

申请日：2015-03-20

申请人： Fraunhofer Gesellschaft zur Förderung der angewandten Forschung E.V. ,  Technische Universität Berlin

发明人： LAPUSCHKIN, Sebastian ,  SAMEK, Wojciech ,  MÜLLER, Klaus-Robert ,  BINDER, Alexander ,  MONTAVON, Grégoire

IPC分类号： G06K9/00 ,  G06T1/20 ,  G06N3/04 ,  G06N3/08

CPC分类号： G06N3/02 ,  G06F17/2765 ,  G06K9/4628 ,  G06K9/6247 ,  G06N3/04 ,  G06N3/084 ,  G10L25/30

摘要： The task of relevance score assignment to a set of items onto which an artificial neural network is applied is obtained by redistributing an initial relevance score derived from the network output, onto the set of items by reversely propagating the initial relevance score through the artificial neural network so as to obtain a relevance score for each item. In particular, this reverse propagation is applicable to a broader set of artificial neural networks and/or at lower computational efforts by performing same in a manner so that for each neuron, preliminarily redistributed relevance scores of a set of downstream neighbor neurons of the respective neuron are distributed on a set of upstream neighbor neurons of the respective neuron according to a distribution function.