Systematic’s data specialists identify suspicious ships and assist home care assistants
Among the thousand or so coders at Systematic sits a small eight-strong team that is building the foundation for the company’s future. Systematic’s Data Science group is digging for gold in the data that will carry the company forward.
Two of Systematic’s data science specialists: Peter Hinge, a specialist in machine learning, and Maiken Gustafsson, who holds a PhD in astrophysics. The Data Science group functions as an internal ‘consultancy’, solving assignments across the organisation. In-house, it is referred to as a centre of excellence, and their specialist know-how is in high demand. Photo: Kåre Viemose
They are constantly experimenting and testing out new ideas, with the full support of the senior management team. Systematic’s Data Science group is a highly specialised team with an unusually varied make-up.
It includes, for example, a member who holds a PhD in astrophysics, an economist, an analytical philosopher, a computer scientist and another member who holds an MSc in Cognitive Science. In a software company with a preponderance of programmers, the Data Science group is something of a patchwork elephant.
In fact, patchwork elephants can be found throughout Systematic. The Data Science group functions as an internal ‘consultancy’, solving assignments across the organisation. In-house, it is referred to as a centre of excellence, and their specialist know-how is in high demand.
Peter Hinge is a machine learning specialist. Peter joined the Data Science group on the back of his thesis in analytical philosophy combined with a passionate interest in artificial intelligence. Photo: Kåre Viemose
Success extends beyond company’s boundaries
The group receives far more in-house tasks from Systematic’s Defence and Healthcare departments than it has the manpower to solve. In addition, it receives enquiries from the company's customers who need help with making the most of their data.
The workload means that Peter Hinge, a specialist in machine learning, and Maiken Gustafsson, who holds a PhD in astrophysics, and the rest of the Data Science group need to take on more people to keep pace with demand. The senior management team is fully supportive, and always expresses its confidence in the slightly oddball group.
For Systematic, the Data Science group is a strategic investment in the future. It’s a question of evolving from a ‘traditional’ software company to being data-driven in its strategic and operational decision-making. Tomorrow’s systems must be data-based, with data continually being used to improve the solutions.
The oblique approach is the right approach
“We’re looking for bright people with experience within machine learning and artificial intelligence. Your educational background is secondary, just as long as you are passionate about data science and keen to acquire the required level of domain knowledge,” says Peter Hinge, a specialist in machine learning. Peter joined the department after completing a thesis in analytical philosophy combined with a passionate interest in artificial intelligence.
Maiken Gustafsson, who holds a PhD in astrophysics, and has 12 years of experience with data science and business intelligence from MHI Vestas under her belt, adds:
“We don’t follow set formulas or templates in our work – we experiment and try things out. Therefore, we’re looking for people who can handle a dynamic working day with lots of specialist challenges and the freedom to solve assignments in new ways.”
Maiken Gustafsson, who holds a PhD in astrophysics, is a new employee who joined the group after 12 years working with data science and business intelligence at MHI Vestas. Photo: Kåre Viemose
Solutions for home care and monitoring offshore waters
The Data Science group has developed the first version of an anomaly detection system for monitoring offshore waters. The system uses deep learning and graph analytics to spot ships taking suspicious routes or which are trying to conceal their identity. The system can help reveal human trafficking, exchanges of smuggled goods and the illegal dumping of oil.
Another project involves developing Cura, Systematic’s software system for home care. A new interface will make it easier for care staff to interact with the system by means of speech. Voice control of the software and the possibility of reading text aloud facilitates interaction. Natural-language understanding (NLU), bot technology, text-to-speech and speech-to-text are some of the technologies which the Data Science group is using in the project.
Classic approach vs deep learning
The group solves many of its tasks by employing a classic data science approach. One sets out baseline models with different regressions, classifications and boosting and trees methods, and then the group tries to beat them with deep learning.
“Basically, we’re deep learning fans and crazy about deep neural networks. If you’ve got the right data material, then it’s usually deep learning that is best at solving the task. If not, we simply add some more layers,” says Peter Hinge, smiling.
According to Maiken Gustafsson, being a data scientist at Systematic offers unique possibilities:
“It’s all about data. At Systematic we have two large domains – Healthcare and Defence – and therefore access to large volumes of data. We have the databases for most of the hospitals in Denmark, which gives us a unique starting point for designing advanced solutions. Now we just need a few more colleagues who are passionate about data science to join us so that we can solve all the exciting tasks that are waiting for us out there.”