Examining Redundancy in the Context of Safe Machine Learning

TL;DR

This paper investigates the challenges of using redundant neural network architectures for safe machine learning by conducting experiments on MNIST, highlighting potential difficulties in deploying neural networks in dependable systems.

Contribution

It provides empirical insights into the limitations of naive redundancy approaches in neural networks for safety-critical applications.

Findings

Redundant architectures face significant challenges in safety-critical contexts.

Experiments reveal difficulties in ensuring dependability with naive neural network redundancies.

Results underline the need for more robust methods for safe machine learning.

Abstract

This paper describes a set of experiments with neural network classifiers on the MNIST database of digits. The purpose is to investigate na\"ive implementations of redundant architectures as a first step towards safe and dependable machine learning. We report on a set of measurements using the MNIST database which ultimately serve to underline the expected difficulties in using NN classifiers in safe and dependable systems.