Felipe Tobar

About

I am an Associate Professor in Machine Learning in the Department of Mathematics at Imperial College London. My research develops probabilistic machine learning methods for modelling complex temporal data, combining advances in statistical signal processing, Bayesian inference, and modern generative modelling. I have applied these methods to problems in the life sciences, astronomy, and audio processing. Alongside my research, I have led institutional initiatives that strengthen collaboration between mathematics, industry, the public sector, and other scientific disciplines at both Imperial College London and Universidad de Chile. I have also contributed extensively to graduate education through curriculum design, programme leadership, and postgraduate supervision.

Prospective PhD students: I am looking for strong and motivated students to join my group at Imperial. If you are interested, please get in touch and apply through the Department of Mathematics, or the Centres for Doctoral Training CCMI and StatML.

E-mail: first initial (dot) last name (at) imperial (dot) ac (dot) uk

News (recent & upcoming)

1 June 2026 I will be giving a tutorial at the upcoming 2nd Spring School on Deep Learning in Buenos Aires, Argentina. Registration deadline is 30 June.
18 Jan 2026 We have three papers accepted at ICASSP 2026, see you in Barcelona! Preprints:
- Unconditional flow-based time series generation with equivariance-regularised latent spaces (link)
- Accelerated training of Gaussian processes using banded square exponential covariances (link)
- Efficient Gaussian process learning via subspace projections (link)
17 Dec 2025 Our paper “Diffusion Self-Weighted Guidance for Offline Reinforcement Learning” has been accepted at TMLR. Link here.
16 Dec 2025 I am giving an invited talk on the 2025 IMS International Conference on Statistics and Data Science in Seville titled Training Gaussian processes beyond maximum likelihood: Distances, approximations, and projections.
5 Nov 2025 I am very happy to have contributed with a chapter for the report AI for a Planet Under Pressure. Available Online.
1 October 2025 Welcome Skye Purchase! Skye is joining my group as a PhD student in the CDT CCMI, and will be working on physics-aligned generative modelling.
1 October 2025 This year I will be teaching two modules at Imperial. The first one is Mathematics for Machine Learning (w/ Eleonora Giunchiglia) in the MSc in Artificial Intelligence Applications and Innovation. The second one is Probabilistic Generatives Models in the MSc in Applied Mathematics. Lecture notes and other material will be up soon.
1 August 2025 Our work “Using language models to detect and reduce gender bias in university forum messages” has been accepted at IEEE Signal Processing Magazine. Available here
4 July 2025 I am giving a talk on Generative Models for Time Series at IRIT. I am very happy to visit my colleagues in Toulouse!

Research group

Current

Camilo Carvajal (PhD in Mathematics, Imperial): Flow models for time series
Skye Purchase (PhD in Mathematics, Imperial): Physics-aligned generative modelling
Florent Yalinkpon (visiting from Université d’Abomey-Calavi, Benin): Prediction of agricultural variables via domain adaptation (Mar–July, 2026)

Former Postdoc

Elsa Cazelles, now CNRS Researcher at IRIT

Former PhD student

Gonzalo Ríos, now VP of Marketing at Niuro

Former Data Scientists

Taco de Wolff, Founder DeWolff Consulting
Bryan Sagredo, Senior Machine Learning Engineer, Marker Learning

Former MSc (research) students

Imperial, MSc in Applied Mathematics: Emily Ehrhardt, Aryan Kumar, Arkin Fernandes.
Universidad de Chile, Master in Data Science: Fernando Fetis, Joaquín Barceló, Camilo Carvajal, Diego Cortéz, Gianina Salomó, Stefano Schiappacasse, Cristóbal Alcázar,
Universidad de Chile, MSc in Applied Mathematics: Tomás Valencia, Bruno Moreno, Sebastián López, Matías Altamirano, Alejandro Cuevas, Mauricio Campos, Rodrigo Lara, Gabriel Parra
Universidad de Chile, MSc in Electrical Engineering: Jou Hui Ho, Lerko Araya, Iván Castro,

Short Bio

Felipe Tobar is an Associate Professor in Machine Learning at the Department of Mathematics, at Imperial College London. Previously, he was an Associate Professor at Universidad de Chile and the Director of the Initiative for Data and Artificial Intelligence of the same Institution. He is an invited researcher at the Center for Mathematical Modeling and the Advanced Center for Electrical and Electronic Engineering. Felipe was a postdoc at the Machine Learning Group, University of Cambridge, during 2015 and received a PhD in Signal Processing from Imperial College London in 2014. Felipe's research interests lie in the interface between Machine Learning and Statistical Signal Processing, including approximate inference, Bayesian nonparametrics, spectral estimation, optimal transport and Gaussian processes.

Photos: Color, BW.