By Dan Horner
Since 2016, the HacktheMachine competition has challenged software engineers, data scientists, government and military personnel, graphic designers, students, and start-up CEOs to solve some of the Navy’s toughest digital problems. I attended this year’s event in New York City, where I had the opportunity collaborate with a wide variety of technologists and tackle challenges directly related to my work at Veracity. I certainly learned a great deal.
The event’s objective was to build “a community of practice for maritime cyber,” and it had a $90,000 purse of prize money. Previously, the competition focused on cyber operations, where the participants attempted a number of hacking tasks against simulated shipboard systems (hence the moniker “HacktheMachine”). The 2019 event introduced two additional competition tracks: creating tools and approaches to enable remote prototyping of 3-D-printed repair parts and developing data-science and machine-learning techniques to discern patterns and anticipate maintenance issues for F/A-18 aircraft. Because of my experience performing predictive analysis and forecasting for naval aviation inventory management, I naturally gravitated to the F/A-18 data-science challenge.
The event kicked off Friday evening aboard the USS Intrepid Sea, Air, and Space Museum in Manhattan, with remarks from the Navy’s Chief Engineer, RDML Selby (SEA-05: Ship Design, Integration, and Engineering) and the head of Naval Information Warfare Systems Command, RADM Becker.
The attendees then dispersed into introductory sessions for each of the three challenge areas. In the F/A-18 data science track, the overall challenge was to “identify patterns and correlations in the aircraft memory-unit (MU) data and determine how those relate to maintenance actions.” We had two primary tasks to perform, using approximately 7 years of real-world data from 45 F/A-18 aircraft:
- Identify which 13 of the 45 aircraft had a particular wiring inspection and subsequent repairs. Then list the maintenance actions that we believe fixed the wiring issues.
- Develop algorithms, techniques, and visualizations that most effectively identify and demonstrate patterns in the relationship between maintenance service panel (MSP) codes and maintenance actions, particularly those related to corrosion.
I attended the event independently, so I was assigned to a team of eight other solo participants. We reassembled the next morning at 0900 in the New Lab at the Brooklyn Navy Yard. The work venue was one large open space, facilitating a great commotion of idea-generation, direct collaboration, and creative expression. There we set to work until Sunday afternoon, when the judging would take place.
The F/A-18 problem-sets were quite difficult; the first was like looking for 13 needles in a haystack—without knowing exactly what those needles should look like. The second was extremely open-ended, allowing for an array of approaches. Though many of my teammates had little experience in this realm, by Sunday morning we had devised a reasonable algorithm for the first challenge and a couple of methods to resolve the second. We then produced the code and documentation that were required for submission, and I delivered a short presentation of our work to the judges.
My team did not end up winning any of the prizes, but in the future, if several of us from Veracity attended as a team, our combined expertise would make us quite competitive. Events like HacktheMachine give data scientists like us the opportunity put our knowledge and creativity to the test, working on operational problems with real-world data, while allowing the Navy to explore a variety of promising solutions. Indeed, this type of work has great potential to benefit the fleet, and teams participating this year barely scratched the surface for developing and applying this type of analysis. The Navy truly presented us with some of its toughest challenges, and it is satisfying to support its operations with data science and machine learning, not only in competitions like this, but in my daily work.