The international discussion on military applications of machine learning has focused much more on algorithmic decision-making than on data. Yet without data, an algorithm cannot be trained, and even after training, an algorithm requires data in its processing to make decisions.
This primer explores data in the context of military decision-support tools and increasingly autonomous weapons systems by briefly discussing the links of the data chain (creation, collection, organization and use), potential challenges to data integrity in adversarial environments, and concludes with a few forward-looking questions for policymakers considering military applications of increasingly autonomous systems.
Teaser: A Primer
Citation: Lydia Kostopoulos (2019) "The Role of Data in Algorithmic Decision-Making", UNIDIR, Geneva, Switzerland.