Man sitting at laptop

Department 70 researchers present paper on Self-Explaining Machine Learning at Digital Avionics Systems Conference


Department 70 researchers Alex Stringer, Brian Sun and Jenniffier Bieberbach will present a paper at the Digital Avionics Systems Conference in October. The paper is focused on the realm of explainable Machine Learning, specifically for radar and command and control applications for Department of Defense aircraft.

Department 70 is the Oklahoma City Air Logistic Complex’s (OC-ALC) applied research laboratory. They focus on research that is relevant to the OC-ALC and supports the various platforms that it sustains. The Digital Avionics Systems Conference (DASC) is an academic conference backed by the Institute of Electrical and Electronics Engineers (IEEE) and their focus in on sharing research in the realm of digital avionics. This conference gives academics, industry, and government researchers the opportunities to show what they are working on and present their research findings. Sharing knowledge and ideas in this capacity can advance the field and make avionics systems more capable: increasing performance and introducing new capabilities that nobody thought were possible. The 76th Software Engineering Group and the OC-ALC are collaborating with OU and the Air Force Research Labs (AFRL) on their Machine Learning research. The AFRL focuses on basic, fundamental research and partners with groups like the OC- ALC and academic partners who can help bring that fundamental research to an applied, practical setting. Department 70 is using some of the Machine Learning tools that the AFRL has developed to create the algorithm that they are developing with professors at OU. The research that Department 70 does with regards to self-explaining Machine Learning is imperative to the technological advancement of today’s warfighters. Currently, one of the key limitations to the broad incorporation of Machine Learning in the defense sector is that the end user has no way of interpreting the decisions that the system makes. In mission critical and safety critical environments, like many of those encountered in the aerospace and defense field, it is important for the people involved to know why a decision gets made. Those in charge of command and control operations want to know how the aircraft are being flown, why they are making turns when they do, and how they will avoid dangerous situations. Machine Learning algorithms are traditionally black boxes; when they learn to make decisions you don’t know why they are making those decisions. Department 70’s focus is on connecting those dots by developing a Machine Learning algorithm that can explain itself. It can generate explanations of its own decisions in a way that is easy to understand for an average human being. In an operational setting for the warfighter, that means that the aircraft may be able one day to pilot itself and report on its actions or a Machine

Learning algorithm could help analyze large amounts of situational data to identify important things in the environment and explain why they are important. Machine Learning can perform these tasks far more quickly and accurately than a person, which can be the difference between a successful mission and a disaster. Machine Learning is the focus of the upcoming Digital Avionics Systems Conference, and the Department 70 research team will be one of four groups presenting on the topic of self-explaining Machine Learning as it relates to avionics. The 76th Software Engineering Group is extremely proud to have such a bright team of researchers, who love what they do and are passionate about the subjects that they research. Their hard work and dedication are certainly making a positive impact on the technical advances necessary to keep the United States Air Force the best in the world.


AFSC Shield AFSC Software Directorate
AFSC.SWEG.SW@us.af.mil