Presentation Type
Lecture

Cognitive EW: Data Requirements for AI

Presenter
Title

Karen Haigh

Country
USA
Affiliation
Haskill Consulting

Presentation Menu

Abstract

A common myth is that we need a lot of data to train an AI-ML system. This myth not only causes developers to fear data collection, but also causes developers to create solutions that are not fieldable. EW systems can leverage first-principle models (such as physics), feature engineering, and progression of the engagement to dramatically reduce the data collection requirements, and create systems that can learn on single observations of novel environments.