Guiomar Pescador Barrios

Hello! I am a final year PhD candidate in Statistics and Machine Learning at Imperial College London and the University of Oxford, where I am fortunate to be advised by Sarah Filippi and Mark van der Wilk.

Before starting my PhD, I completed a MMath Mathematics at the University of Edinburgh (2017-2022).

I am interested in developing variational methods that support continual and adaptive learning. One line of my work focuses on methods that automatically adjust the model’s size as learning progresses, reducing computational cost while maintaining accuracy. Recently, I have been working on optimisation methods.

news

Dec 02, 2025	Attending EurIPS 2025 in Copenhagen 🇩🇰
Nov 24, 2025	I’ll be presenting at Breaking Topics in AI Conference 2025 in London.
May 03, 2025	Excited to announce that our paper on adaptive model size for continual Gaussian Processes has been accepted as a Spotlight at ICML 2025! Also received the G-Research Early-Career Research Grant to attend the conference in Vancouver, Canada. 🎉
Apr 03, 2025	Began my academic placement at RIKEN Center for Advanced Intelligence Project in Tokyo, Japan! Working with Dr. Emtiyaz Khan’s Adaptive Bayesian Intelligence Team.
Dec 10, 2024	Lightning talk at NeurIPS 2024 Bayesian Decision-making and Uncertainty Workshop on adaptive model size for continual Gaussian Processes.
Nov 18, 2024	Attending Dagstuhl Seminar on “Rethinking the Role of Bayesianism in the Age of Modern AI” in Germany.

selected publications

ICML

Adjusting Model Size in Continual Gaussian Processes: How Big is Big Enough?

Guiomar Pescador-Barrios, Sarah Filippi, and Mark van der Wilk

In Forty-second International Conference on Machine Learning (ICML), 2025

SPOTLIGHT Bib PDF Code

Spotlight Paper (top 2.6% of all submissions)

@inproceedings{pescador-barrios2025adjusting,
  title = {Adjusting Model Size in Continual Gaussian Processes: How Big is Big Enough?},
  author = {Pescador-Barrios, Guiomar and Filippi, Sarah and Wilk, {Mark van der}},
  booktitle = {Forty-second International Conference on Machine Learning (ICML)},
  year = {2025},
}