Christopher Sprague

Christopher's research mainly revolves around combining learning- and model-based methods to enhance the application of artificial intelligence.

Currently, he is working as a postdoc with Hossein Azizpour and Arne Elofsson at SciLifeLab, researching how to enhance the modelling of protein interaction with deep learning techniques. This research area includes, for example, protein-protein interaction, drug discovery, and protein folding. From a machine learning perspective, this research area is distinct for the reason that it is situated in a relatively low-data regime and there are a plethora of geometry- and physics-based inductive biases that can be leveraged to enhance the performance of machine learning models.

Christopher earned a PhD in computer science from KTH Royal Institute of Technology in 2022, working with Petter Ögren. His doctoral research focused on efficient and trustworthy artificial intelligence for critical robotics systems (thesis). These systems are characterised by the need for both efficiency (the robot does what it's supposed to do in a timely manner) and safety (the robot doesn't do what it's not supposed to do). While learning-based methods can encourage efficiency and safety, they often do not guarantee it. On the other hand, while model-based methods may not be as efficient as learning-based methods, they often provide formal guarantees of efficiency and safety in the form of convergence (doing what is supposed to be done) and set invariance (not doing what is not supposed to be done), which provides trustworthiness. Christopher's research explored how efficiency and safety could be enhanced by learning-based methods, while at the same time guaranteed by model-based methods.

Before his PhD, Christopher earned a BS in 2016 and an MS in 2017 in aerospace engineering at Rensselaer Polytechnic Institute, where he was advised by Kurt Anderson and studied solid mechanics, fluid mechanics, and dynamical systems. In 2017, he worked with Dario Izzo at the European Space Agency's Advanced Concepts Team, researching deep learning for interplanetary spacecraft trajectory design. In 2017, he won the East National Science Foundation (NSF) Asia and Pacific Summer Institutes for U.S. Graduate Students (EAPSI) fellowship and worked with Yasuhiro Kawakatsu at the Japan Aerospace Exploration Agency, also focusing on deep learning for trajectory design. In 2015, he had an internship with NASA, working at the Johns Hopkins Applied Physics Laboratory on the fault protection systems of the STEREO spacecraft.

Aside from research, Christopher competed in athletics and cross country in both high school and college, focusing on 800m to 1500m distances.