Characterizing the Decision Boundary of Deep Neural Networks

TL;DR

This paper introduces DeepDIG, a novel method for generating samples near the decision boundaries of deep neural networks using adversarial examples, enabling better understanding of their decision-making behavior.

Contribution

The paper presents DeepDIG, a new approach for characterizing decision boundaries of neural networks through adversarial instance generation and principled analysis.

Findings

DeepDIG effectively generates boundary samples across various models.

Analysis reveals key properties of neural network decision boundaries.

Extensive experiments validate the approach's utility.

Abstract

Deep neural networks and in particular, deep neural classifiers have become an integral part of many modern applications. Despite their practical success, we still have limited knowledge of how they work and the demand for such an understanding is evergrowing. In this regard, one crucial aspect of deep neural network classifiers that can help us deepen our knowledge about their decision-making behavior is to investigate their decision boundaries. Nevertheless, this is contingent upon having access to samples populating the areas near the decision boundary. To achieve this, we propose a novel approach we call Deep Decision boundary Instance Generation (DeepDIG). DeepDIG utilizes a method based on adversarial example generation as an effective way of generating samples near the decision boundary of any deep neural network model. Then, we introduce a set of important principled characteristics that take advantage of the generated instances near the decision boundary to provide multifaceted understandings of deep neural networks. We have performed extensive experiments on multiple representative datasets across various deep neural network models and characterized their decision boundaries. The code is publicly available at https://github.com/hamidkarimi/DeepDIG/.