Layer-Wise Relevance Propagation for Deep Neural Network Architectures

By Alexander Binder, Sebastian Bach, Grgoire Montavon, K. Müller, Wojciech Samek. 2016

173 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Explanation Methods in Deep Learning: Users, Values, Concerns and Challenges (2018, background): “Attribution, a term introduced by [2], also referred to as relevance [5, 7, 60, 47], contribution [51], class saliency [52] or influence [27, 1, 28], aims to reveal components of high importance in the input to the DNN and their effect as the input is propagated through the network.”
Attacks on Machine Learning: Lurking Danger for Accountability (2019, background): “Nevertheless, we believe that a new field of attacks on ML components will open up in this field in the future because initiatives such as eXplainable AI (layer-wise relevance propagation [16], Black Box Explanations through Transparent Approximations (BETA) [37], LIME [58], Generalized Additive Mod”
Deep Learning for Time Series Forecasting: Tutorial and Literature Survey (2020, background): “, [19, 122, 158]) are of particular practical importance in the forecasting setting.”
Explainable AI: Interpreting, Explaining and Visualizing Deep Learning (2019, methodology): “Yet [13] quotes General Michael Hayden, former director of the NSA and CIA, as saying “We kill people based on metadata.”
On internal language representations in deep learning: an analysis of machine translation and speech recognition (2018, methodology): “For instance, layer-wise relevance propagation (LRP), which propagates a measure of relevance down the network [37] has been applied to several language processing tasks [10, 11], including neural machine translation [90].”
VisualBackProp: efficient visualization of CNNs (2018, methodology): “The stability of the method and the sensitivity to different settings of the conservation parameters was studied in the context of several deep learning models [12].”
VisualBackProp: visualizing CNNs for autonomous driving (2016, background): “Its stability and the sensitivity to different settings of the conservation parameters was studied in the context of several deep learning models (Binder et al., 2016).”
Under the Hood of Neural Networks: Characterizing Learned Representations by Functional Neuron Populations and Network Ablations (2020, methodology): “…a network include methods like gradient based class activation mapping (Grad-CAM) (Selvaraju et al., 2017), layer-wise relevance propagation (LRP) (Binder et al., 2016), deep taylor decomposition (Montavon et al., 2016) and class-enhanced attentive response (CLEAR) (Kumar et al., 2017), which…”
Towards Explainable Artificial Intelligence (2019, methodology): “Layer-wise Relevance Propagation (LRP) [9,58] is a propagation-based explanation framework, which is applicable to general neural network structures, including deep neural networks [13], LSTMs [7,5], and Fisher Vector classifiers [44].”