Hi, my name is

Homer Durand.

What are probabilities? Is the idea of causality necessarily attached to the concept of counterfactuals? Can we assess blame and responsibility without counterfactual reasoning? Here are some of the questions I am currently very interested about.

About Me

I amm a PhD student at the Universitat de València (Spain), supervised by Gustau Camps-Valls and Gherardo Varando. My research focuses on developing and understanding Detection and Attribution (D&A) of Climate Change methods in a causal inference framework.

More broadly, I’m interested in learning theory and how to combine different sources of data and knowledge to make predictions and understand complex phenomena. My academic interests span causal inference, Bayesian inference (especially as inductive logic), kernel methods, climate change D&A, the philosophy of causality and probability, and moral philosophy—particularly how blame and responsibility relate to counterfactual reasoning.

I recently had a paper accepted at AISTATS on causal regularisation for robust multivariate prediction, and another submission on a causal representation learning framework for identifying direct effect spaces in outcome variables, with applications to detection and effect separation.

I am also preparing a course on probabilistic climate modelling. It will explore probability as inductive logic, uncertainty quantification, Markov Chain simulations in climate models, calibration and basic concepts in detection and attribution.

Experience

PhD student researcher - ISP-IPL lab
november 2022 - present
Currently doing my PhD thesis in Remote Sensing at Universitat de València and working as a researcher at the Image Processing Lab in the Image and Signal Processing group. My work focus on developping Machine Learning methods based on causality, kernel methods and dimensionality reduction for the Detection and Attribution of Climate Change.
Intern Researcher - ISP-IPL lab
June 2022 - Oct 2022
I worked as an intern researcher at Image Processing Lab (IPL) in the Image and Signal Processing (ISP) group. My research topic was Causal Representation Lerning for Earth System Modelling.
Intern Researcher - LOCEAN-IPSL lab
Mar 2021 - Sept 2021
I worked as an intern researcher at Pierre Simon Laplace institute in the LOCEAN group. My research topic was Dynamical model calibration with History Matching methodology using Gaussian Process, Random Forest and Bayesian Neural Network regressions. I mainly worked on assessing wether machine learning algorithms could replace the Gaussian Processes used in History Matching, a parametrization methodology for simulation models.
Intern Software Developer - Kyntus
Jul 2019 - Aug 2019
I worked on the development of a web app for schedules and projects management using PHP, SQL and HTML.

Education

2022 - 2025
PhD in Remote Sensing
Universitat de València, Image Processing Lab, València
  • Thesis: Causal Represntation Learning for Detection and Attribution of Climate Change.

  • Project: Learning causal representation and develop causal tools robust to distributional shifts for spatio-temporal and potentially high-dimensional data. Apply these methodologies to Detection and Attribution of climate change.

  • Advisors: Gustau Camps-Valls, Gherardo Varando

2021 - 2022
Master's degree in Applied Mathematics - specialisation in Statistics
Sorbonne University, Paris
Courses: Nonparametric inference, High dimensional linear models, Statistical learning, Sequential convex optimization, Bayesian statistics, Statistical analysis of graphs, Statistical models for ecology.
2018 - 2021
Master's degree in Engineering - Specialisation in Applied Mathematics and Computer Science
Polytech Sorbonne, Paris
Courses: Functional analysis, Numerical analysis, Probability, Statistics, Machine Learning, Data Analysis, Convex and non-convex optimization, High Performance Computing, Linear Algebra, Mesure Theory

Projects

Out-of-distribution Generalisation via Diluted Causality
Out-of-distribution Generalisation via Diluted …

📄 Paper accepted at AISTATS 2025

Description

The term diluted causality comes from Bühlmann (2018), …

History Matching for climate model tuning - experiments on the Lorenz96 toy model
History Matching for climate model tuning - …

Description

During my end-of-study internship at the LOCEAN-IPSL laboratory, I have been working on …

Machine Learning for the detection of Deficient MMR Crypts to aid in the diagnosis of Lynch Disease
Machine Learning for the detection of Deficient …

Description

Lynch Syndrome is associated with a significantly elevated risk of developing …