Magnus Ross

Hi! I’m Magnus and this is my website, hopefully you will find something interesting here. I am currently a postdoctoral research fellow based in the AI Center at UCL. I’m interested in forecasting, uncertainty quantification, and probabilistic ML in general. I am also working on using ML to help combat antimicrobial resistance, as part of the Digital Health Hub for AMR.

Previously, I did my PhD at the University of Manchester working with Mauricio Alvarez on integrating physical information into Gaussian process models. You can find my thesis here. For a time I worked for the dark side (a trading firm), building forecasting models for power and other energy commodities.

Outside of work I like to go running, play music, and sit in the sauna.

Some useful links:

A panoramic view of mountains on a clear day. The valley below is coverd by fog. — A nice view from near my home town.

Posts

2024/11/29 Time series benchmarks are broken
2023/06/19 Expectation maximization for mixture of linear experts
2023/04/20 Distributions of quotients of uniform RVs
2023/04/12 Mercer theorem measure and RKHS norm
2023/04/03 Converting state misalignment in probabilistic numeric ODE solver to SDE.
2023/03/22 Relationship between the Frobenius norm and singular values in SVD
2023/03/14 New paper: Learning Energy Conserving Dynamics Efficiently with Hamiltonian Gaussian Processess
2023/03/14 Winners curse in Bayesian optimisation
2021/07/15 New paper: Learning Nonparametric Volterra Kernels with Gaussian Processes
2021/01/17 Simulation is easy, probability is hard…
2021/01/04 Talk at 3rd Sheffield Workshop on Structural Dynamics
2020/10/03 Podcasts about ML
2020/07/22 Lagrangians with PyTorch Part #1