Pixel-Wise Explanations for Non-Linear Classifier Decisions
Failed to add items
Add to basket failed.
Add to Wish List failed.
Remove from Wish List failed.
Follow podcast failed
Unfollow podcast failed
Pixel-Wise Explanations for Non-Linear Classifier Decisions
-
Narrated by:
-
By:
About this listen
This open-access research article from PLOS One introduces Layer-wise Relevance Propagation (LRP), a novel method for interpreting decisions made by complex, non-linear image classifiers. The authors, an international team of researchers, explain how LRP can decompose a classification decision down to the individual pixels of an input image, generating a heatmap that visualizes their contribution. This technique aims to make "black box" machine learning models, like neural networks and Bag of Words (BoW) models, more transparent by showing why a system arrives at a particular classification. The paper evaluates LRP on various datasets, including PASCAL VOC images and MNIST handwritten digits, and contrasts it with Taylor-type decomposition, providing a comprehensive framework for understanding and verifying automated image classification.
Source: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0130140