Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Layer-wise relevance propagation for neural networks with local renormalization layers

: Binder, A.; Montavon, G.; Lapuschkin, S.; Müller, K.-R.; Samek, W.


Villa, A.E.P.:
Artificial neural networks and machine learning - ICANN 2016. Proceedings Part 2 : 25th International Conference on Artificial Networks, Barcelona, Spain, September 6-9, 2016
Berlin: Springer, 2016 (Lecture Notes in Computer Science (LNCS) 9887)
ISBN: 978-3-319-44780-3
ISBN: 978-3-319-44781-0
International Conference on Artificial Neural Networks (ICANN) <25, 2016, Barcelona>
Fraunhofer HHI ()

Layer-wise relevance propagation is a framework which allows to decompose the prediction of a deep neural network computed over a sample, e.g. an image, down to relevance scores for the single input dimensions of the sample such as subpixels of an image. While this approach can be applied directly to generalized linear mappings, product type non-linearities are not covered. This paper proposes an approach to extend layer-wise relevance propagation to neural networks with local renormalization layers, which is a very common product-type non-linearity in convolutional neural networks. We evaluate the proposed method for local renormalization layers on the CIFAR-10, Imagenet and MIT Places datasets.