Layer-Wise Relevance Propagation: An Overview

By Grgoire Montavon, Alexander Binder, Sebastian Lapuschkin, Wojciech Samek, Klaus‐Robert Müller . 2019

897 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biophysics, Safety, Robustness & Interpretability

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 Biophysics

In Safety, Robustness & Interpretability

Citations in the corpus

Unbox the black-box for the medical explainable AI via multi-modal and multi-centre data fusion: A mini-review, two showcases and beyond (2021, methodology): “[158] tested an XAI method using Layer-wise Relevance Propagation [159] for the prediction of acute critical illness from EHR.”
Modern Hopfield Networks and Attention for Immune Repertoire Classification (2020, neutral): “, 2017) or Layer-Wise Relevance Propagation (Montavon et al., 2018; Arras et al., 2019; Montavon et al., 2019; Preuer et al., 2019).”
Towards Explainable Artificial Intelligence (2019, background): “Layer-wise Relevance Propagation (LRP) [9,58] provides a general framework for explaining individual predictions, i.”
Deep Neural Networks and Tabular Data: A Survey (2021, neutral): “For such settings, relevance-propagation-based methods, e.g., [161], [162], and gradient-based approaches, e.g., [163], [164], [165], have been suggested.”
Explainable AI: Interpreting, Explaining and Visualizing Deep Learning (2019, methodology): “Layer-wise Relevance Propagation (LRP) [9,58] provides a general framework for explaining individual predictions, i.”
Toward Interpretable Machine Learning: Transparent Deep Neural Networks and Beyond (2020, methodology): “The latter is indeed speaking for the class locomotive, which mechanistically lowers the probability for the class ‘passenger car’ [98].”
Proactive Pseudo-Intervention: Causally Informed Contrastive Learning For Interpretable Vision Models (2020, methodology): “Further, the chosen LRP models, which focus on either higher relevance points ( LRP (cid:15) , LRP w 2 ) or positively weighted relevance points ( LRP γ , LRP αβ ) [39], give very sharp saliency maps.”
Explainability for Large Language Models: A Survey (2023, methodology): “Layer-wise relevance propagation ( LRP ) [Montavon et al. 2019] and Taylor-type decomposition ( DTD ) approaches [Montavon et al. 2015] are two classes of commonly used methods.”
Higher-Order Explanations of Graph Neural Networks via Relevant Walks (2020, methodology): “cting quantities in the ML model) have received growing attention. Second-order methods (e.g. based on the model's Hessian) have been proposed to attribute predictions to pairs of input features [9, 11, 8]. Other methods have been proposed to identify the collective eect of larger groups of var”
Structured Pruning Adapters (2022, methodology): “CIFAR-10 accuracy versus total learned parameter count (cid:107) ∆ W t (cid:107) 0 (top row) and FLOPs (bottom row) using ﬁne-pruning and SPLoRA with ranks 32 and 8 for the (a) Weight (Li et al., 2017), (b) Gradient (Sun et al., 2017), (c) Taylor (Molchanov et al., 2017), and (d) LRP (Yeom et al., 2”
Human-in-the-loop Debugging Deep Text Classifiers (2020, background): “For more details about LRP propagation rules, please see Montavon et al. (2019).”
Analyzing ImageNet with Spectral Relevance Analysis: Towards ImageNet un-Hans'ed (2019, methodology): “Following [7], we provide attributions computed with Layer-wise Rel-evance Propagation (LRP) [10] according to the recommended composite (or layer-dependent) strategy [23, 24].”
Explainable AI Methods - A Brief Overview (2020, background): “Diﬀerent redistribution rules, adapted to the speciﬁc properties of particular neural network layers, have been proposed for LRP [58, 42].”
Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications (2021, methodology): “A similar procedure with another choice of reference point gives LRP- (see [126]).”