About

I am a research fellow in Molecular Modelling and Probabilistic Machine Learning at the University of Cambridge. My interests span generative models applied to molecular modelling, probabilistic modelling, approximate inference and information theory. My PhD research focused on scalable probabilistic reasoning with neural network models. I graduated from the University of Zaragoza in 2018 with an honorary distinction (“premio extraordinario”) in Telecommunications Engineering (EE/CS). I was awarded an MPhil in Machine Learning with distinction by the University of Cambridge in 2019. I also do freelance engineering consulting and am a co-founder of arisetech.es. Bellow are links to some of my recent work, where ^* denotes equal contribution.

2024

Improving Linear System Solvers for Hyperparameter Optimisation in Iterative Gaussian Processes
J. A. Lin, S. Padhy, B. Mlodozeniec, J. Antorán, J. M. Hernández-Lobato
Preprint, 2024
[Paper]

PhD Thesis: Scalable Bayesian Inference in the Era of Deep Learning: From Gaussian Processes to Deep Neural Networks
Javier Antorán
[Thesis]

A Generative Model of Symmetry Transformations
J. U. Allingham, B. K. Mlodozeniec, S. Padhy, J. Antorán, D. Krueger, R. E. Turner, E. Nalisnick, J. M. Hernández-Lobato
Preprint, 2024
[Paper]

2023

Stochastic Gradient Descent for Gaussian Processes Done Right
J. Antorán^*, J. A. Lin^*, S. Padhy^*, A. Tripp, A. Terenin, C. Szepesvári, J. M. Hernández-Lobato, D. Janz
Twelfth International Conference on Learning Representations (ICLR), 2024
[Paper]

SE(3) Equivariant Augmented Coupling Flows
Laurence I. Midgley^*, Vincent Stimper^*, Javier Antorán^*, Emile Mathieu^*, B. Schölkopf, J. M. Hernández-Lobato
Spotlight at Neural Information Processing Systems (NeurIPS), New Orleans, USA. 2023.
[Paper]

Sampling from Gaussian Process Posteriors using Stochastic Gradient Descent
J. Antorán^*, J. A. Lin^*, S. Padhy^*, D. Janz, J. M. Hernández-Lobato, A. Terenin
Oral presentation at Neural Information Processing Systems (NeurIPS), New Orleans, USA. 2023.
[Paper]

Online Laplace Model Selection Revisited
J. A. Lin, J. Antorán, J. M. Hernández-Lobato
Contributed talk at 5th Symposium on Advances in Approximate Bayesian Inference (AABI). 2023.
[Paper]

Fast and Painless Image Reconstruction in Deep Image Prior Subspaces
R. Barbano, J. Antorán, J. Leuschner, J. M. Hernández-Lobato, Ž. Kereta, B. Jin
11th Applied Inverse Problems Conference (AIP2023)., 2023
[Paper]

Sampling-based inference for large linear models, with application to linearised Laplace
J. Antorán^*, S. Padhy^*, R. Barbano, E. Nalisnick, D. Janz, J. M. Hernández-Lobato
Eleventh International Conference on Learning Representations (ICLR), 2023
[Paper]

2022

Learning Deep Generative Models with Invariance under Symmetry Transformations
J. U. Allingham, Javier Antoran, S. Padhy, E. Nalisnick, J. M. Hernández-Lobato
NeurIPS Workshop on Symmetry and Geometry in Neural Representations, 2022
[Paper]

Bayesian Experimental Design for Computed Tomography with the Linearised Deep Image Prior
R. Barbano^*, J. Leuschner^*, J. Antorán^*, B. Jin, J. M. Hernández-Lobato
Adaptive Experimental Design and Active Learning workshop at ICML, 2022
[Paper]

Adapting the Linearised Laplace Model Evidence for Modern Deep Learning
J. Antorán, D. Janz^*, J.U. Allingham^*, E. Daxberger, R. Barbano, E. Nalisnick, J. M. Hernández-Lobato
39th International Conference on Machine Learning (ICML), 2022
[Paper]

Uncertainty Estimation for Computed Tomography with a Linearised Deep Image Prior
J. Antorán^*, R. Barbano^*, J. Leuschner, J. M. Hernández-Lobato, B. Jin
Transactions on Machine Learning Research 12/2023.
[Paper]

Deep End-to-end Causal Inference
J. Antorán^*, T. G.^*, A. F.^*, W. G., C. M., E. K., A. S., A. L., M. K., N. P., M. A., C. Z.
Project conducted during an internship at Microsoft Research Cambridge.
Arxiv preprint. 2022.
[Paper]

2021

Linearised Laplace Inference in Networks with Normalisation Layers and the Neural g-Prior
J. Antorán, J.U. Allingham, D. Janz, E. Daxberger, E. Nalisnick and J. M. Hernández-Lobato.
Contributed talk at 4th Symposium on Advances in Approximate Bayesian Inference (AABI). 2022.
[Paper], [Poster], [bibtex]

A Probabilistic Deep Image Prior over Image Space
J. Antorán^*, R. Barbano^*, J. M. Hernández-Lobato and B. Jin.
4th Symposium on Advances in Approximate Bayesian Inference (AABI). 2022.
[Paper], [bibtex]

Depth Uncertainty Networks for Active Learning
Chelseay Murray, J. U. Allingham, J. Antorán and J. M. Hernández-Lobato.
Bayesian Deep Learning Workshop at NeurIPS. 2021.
[Paper], [bibtex]

Addressing Bias in Active Learning with Depth Uncertainty Networks… or Not
Chelseay Murray, J. U. Allingham, J. Antorán and J. M. Hernández-Lobato.
Contributed talk at I (Still) Can’t Believe It’s Not Better Workshop at NeurIPS. 2021.
Proceedings of Machine Learning Research, Volume 163. 2022 [Paper], [bibtex]

2020

Bayesian Deep Learning via Subnetwork Inference
E. Daxberger, J. Antorán^*, E. Nalisnick^*, J. U. Allingham^* and J. M. Hernández-Lobato.
Contributed talk at 3rd Symposium on Advances in Approximate Bayesian Inference (AABI), 2020.
38th International Conference on Machine Learning (ICML), 2021.
[Paper], [Poster], [bibtex], [code] [blog]

Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty
U. B., J. Antorán, Y. Z., Q. V. L., P. S., R. F., G. G. M., R. K., J. S., O. T., L. N., R. C., A. W. and A. X.
4th AAAI / ACM conference on Artificial Intelligence, Ethics, and Society (AIES), 2021.
[Paper]

Depth Uncertainty in Neural Networks
J. Antorán, J. U. Allingham and J. M. Hernández-Lobato.
34th Conference on Neural Information Processing Systems (NeurIPS), Vancouver, Canada. 2020.
[Paper], [Poster], [bibtex], [code]

Getting a CLUE: A Method for Explaining Uncertainty Estimates
J. Antorán, U. Bhatt, T. Adel, A. Weller and J. M. Hernández-Lobato.
Oral presentation at The Ninth International Conference on Learning Representations (ICLR), 2021.
[Paper], [Poster], [bibtex], [code]

Variational Depth Search in ResNets
J. Antorán, J. U. Allingham and J. M. Hernández-Lobato.
Contributed talk at 1st Workshop on Neural Architecture Search at ICLR 2020.
[Paper], [Poster], [bibtex], [code]

2019

Understanding Uncertainty in Bayesian Neural Networks
J. Antorán
MPhil Thesis (Awarded Distinction)
[Thesis], [bibtex], [Poster]

Uncertainty in Bayesian Neural Networks (github repo)
J. Antorán and E. Markou.
Presented poster at the Workshop on The Mathematics of Deep Learning and Data Science, The Isaac Newton Institute for Mathematical Sciences, Cambridge, UK. 2019.
[Poster], [Code]

2018

Disentangling in Variational Autoencoders with Natural Clustering
J. Antorán and A. Miguel.
Successfully defended Bachelor’s Thesis at the University of Zaragoza (9.8/10, Honorary Distinction).
Accepted as an oral presentation at the 18th IEEE International Conference on Machine Learning and Applications - ICMLA 2019, Boca Raton, Florida, USA. 2019.
[Paper], [bibtex], [Thesis (Spanish)]

FELIX DAQ Integration Test Tool for the ATLAS experiment at CERN.
J. Antorán and J. Schumacher.
Work done during a Summer studentship at CERN.
[Technical Report], [bibtex]