Adversarial Attack for Uncertainty Estimation: Identifying Critical Regions in Neural Networks

TL;DR

This paper introduces an adversarial attack-based method for neural network uncertainty estimation, effectively identifying critical regions and outperforming existing approaches, especially on blockchain-derived datasets.

Contribution

The paper presents a novel input perturbation technique for uncertainty estimation that differs from Bayesian methods, demonstrating superior performance.

Findings

Significant outperformance over recent uncertainty methods.

Effective identification of decision boundary regions.

Reduced risk in capturing model uncertainty.

Abstract

We propose a novel method to capture data points near decision boundary in neural network that are often referred to a specific type of uncertainty. In our approach, we sought to perform uncertainty estimation based on the idea of adversarial attack method. In this paper, uncertainty estimates are derived from the input perturbations, unlike previous studies that provide perturbations on the model's parameters as in Bayesian approach. We are able to produce uncertainty with couple of perturbations on the inputs. Interestingly, we apply the proposed method to datasets derived from blockchain. We compare the performance of model uncertainty with the most recent uncertainty methods. We show that the proposed method has revealed a significant outperformance over other methods and provided less risk to capture model uncertainty in machine learning.