Layerwise Relevance Visualization in Convolutional Text Graph Classifiers

TL;DR

This paper introduces a novel layerwise relevance visualization method for Graph Convolutional Networks applied to text classification, enabling interpretable insights into intermediate representations and decision-making processes.

Contribution

It presents a new explainability technique that traces features across layers in GCNs, revealing hidden dynamics and improving interpretability of text classifiers.

Findings

Provides meaningful layerwise explanations for GCN sentence classifiers.

Exposes hidden cross-layer dynamics in input graph structures.

Enhances interpretability of deep neural network representations.

Abstract

Representations in the hidden layers of Deep Neural Networks (DNN) are often hard to interpret since it is difficult to project them into an interpretable domain. Graph Convolutional Networks (GCN) allow this projection, but existing explainability methods do not exploit this fact, i.e. do not focus their explanations on intermediate states. In this work, we present a novel method that traces and visualizes features that contribute to a classification decision in the visible and hidden layers of a GCN. Our method exposes hidden cross-layer dynamics in the input graph structure. We experimentally demonstrate that it yields meaningful layerwise explanations for a GCN sentence classifier.