Matt

Background

I completed my master’s degree (MEng) in Engineering Science at the University of Oxford. My master's thesis, supervised by Prof. Stephen Duncan, investigated the stability of pipe flow. In addition to this I completed an internship at the Oxford Man Institute of Quantitative Finance, computing higher order dependencies between features in unstructured financial data sets. I am a student on the AIMS-CDT program and I am now a member of the Control Group, supervised by Prof. Antonis Papachristodoulou, where we have completed two mini-projects on population dynamics and autonomous vehicle control. My current research areas include analysing epidemic models and finding more efficient and accurate methods to verify the robustness of neural networks. I am a college lecturer for Worcester College, where I teach 1st and 2nd year undergraduate students and I have been a departmental teaching assistant for 3rd year undergraduate students.

Research Interests

• Stability of Epidemic Models
• Robustness of Neural Networks and Neural Network Controllers
• Polynomial Optimisation

Publications

1. M. Newton and A. Papachristodoulou, “Neural Network Verification using Polynomial Optimisation,” 2021 60th IEEE Conference on Decision and Control (CDC), Austin, Texas, 2021.
2. M. Newton and A. Papachristodoulou, “Exploiting Sparsity of Neural Network Verification,” 3rd Annual Learning for Dynamics and Control Conference, 2021
3. M. Newton and A. Papachristodoulou, "Network Lyapunov Functions for Epidemic Models," 2020 59th IEEE Conference on Decision and Control (CDC), Jeju, Korea (South), 2020, pp. 1798-1803, doi: 10.1109/CDC42340.2020.9304021.

Other Projects

• An Investigation in to Higher-order Interactions in The Lotka-Volterra Model
• ADMM for Control of Mixed Traffic Flow with Human-driven and Autonomous Vehicles
• The Stability of Plane Poiseuille Flow: A Control Theory Perspective (Master's Thesis)