Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Designing Resilient Manufacturing Systems using Cross Domain Application of Machine Learning Resilience.

In Procedia CIRP

The COVID-19 pandemic and crises like the Ukraine-Russia war have led to numerous restrictions for industrial manufacturing due to interrupted supply chains, staff absences due to illness or quarantine measures, and order situations that changed significantly at short notice. These influences have exposed that it is crucial to address the issue of manufacturing resilience in the context of current disruptions. This can be plausibly guaranteed by subjecting the ML model of a manufacturing system to attacks deliberately designed to fool its prediction. Such attacks can provide useful insights into properties that can increase resilience of manufacturing systems.

Mukherjee Avik, Glatt Moritz, Mustafa Waleed, Kloft Marius, Aurich Jan C

2022

adverserial attacks, adverserial training, deep neural networks, discrete-event simulation environment, machine learning, manufacturing system, resilience, supply network