Bernardin TAMO AMOUGOU

bt2027@hw.ac.uk

I am a dual PhD candidate at Heriot-Watt University and Université Paris Cité, working at the intersection of Bayesian statistics, machine learning, and computational imaging. My doctoral research is supervised by Dr. Andres Almansa, Prof. Julie Delon, and Prof. Marcelo Pereyra.

My research focuses on uncertainty-aware methods for imaging inverse problems, with a particular interest in self-supervised learning, data-driven priors, posterior sampling, and scalable Bayesian computation for medical and scientific imaging.

Before my PhD, I was a research engineer at MAP5, Université Paris Cité, where I worked on stochastic and optimal-transport-inspired algorithms for imaging problems with learned priors. I also completed a Master 2 in Mathematics, Modelling and Machine Learning at Université Paris Cité.

Earlier, I completed the Structured Master’s Degree in Data Science at AIMS Cameroon, where I was the 2021 Valedictorian.

My long-term goal is to build mathematically grounded AI tools for imaging problems where reconstructions inform scientific discovery, medical decision-making, or other high-impact applications.

Research interests: Bayesian imaging, inverse problems, uncertainty quantification, posterior sampling, diffusion and flow models, self-supervised learning, conformal prediction, medical imaging.

Awards and Scholarships

Year	Award
2024	France Excellence EIFFEL Scholarship (French Ministry for Europe and Foreign Affairs)
2023	James Watt Studentship, Heriot-Watt University
2021	SMARTS-UP Scholarship, Université Paris Cité
2021	Valedictorian, AIMS Cameroon Cohort 2021
2021	Gender Balance Prize — Three Minute Thesis, AIMS Cameroon
2020	AIMS Cameroon & Cameroon Government Scholarship (Master’s in Data Science)
2019	Best Student in Mathematical Modelling in Economics and Finance, CETIC
2018	World Bank Scholarship through the African Center of Excellence in ICT (Master 1 & 2 in Computational Finance)

Training Schools

QLA Doctoral Training School — Foundational Methods in Data Science, Quantum Leap Africa, AIMS Rwanda, Kigali, March 2022.
11th Gene Golub SIAM Summer School — The Theory and Practice of Deep Learning, AIMS South Africa, July 2021.

news

May 04, 2026	I presented a poster on equivariant self-supervised VAEs for uncertainty quantification in Bayesian imaging problems at the ICMS workshop in Edinburgh.
May 01, 2026	I applied to participate in the Heriot-Watt University 3 Minute Thesis (3MT) competition, where I will present my research on AI for imaging and decision-making.
Jul 01, 2025	I presented a poster on self-supervised conformal prediction for uncertainty quantification in Poisson imaging at the Machine Learning Summer School (MLSS) in Dakar, Senegal.
Feb 26, 2025	Our preprint on self-supervised conformal prediction for uncertainty quantification in Poisson imaging problems is available online.
Feb 15, 2025	I gave a short talk on self-supervised conformal prediction for uncertainty quantification in imaging at a UCL Workshop in London.

selected publications

Self-supervised conformal prediction for uncertainty quantification in Poisson imaging problems

Bernardin Tamo Amougou , Marcelo Pereyra, and Barbara Pascal

In IEEE Statistical Signal Processing Workshop (SSP) , 2025

Abs arXiv

Image restoration problems are often ill-posed, leading to significant uncertainty in reconstructed images. Accurately quantifying this uncertainty is essential for the reliable interpretation of reconstructed images. This work develops a self-supervised conformal prediction method for Poisson imaging problems, leveraging a Poisson unbiased risk estimator to calibrate uncertainty without ground-truth images. The resulting self-calibrating conformal prediction approach is applicable to any Poisson linear imaging problem that is ill-conditioned, and is particularly effective when combined with modern self-supervised image restoration techniques trained directly on measurement data.
Self-supervised Conformal Prediction for Uncertainty Quantification in Imaging Problems

Jasper M. Everink, Bernardin Tamo Amougou , and Marcelo Pereyra

In International Conference on Scale Space and Variational Methods in Computer Vision (SSVM) , 2025

Abs arXiv

Most image restoration problems are ill-conditioned or ill-posed and hence involve significant uncertainty. This work proposes a self-supervised conformal prediction approach for uncertainty quantification in imaging inverse problems, using Stein’s unbiased risk estimator (SURE) to self-calibrate directly from noisy measurements, bypassing the need for ground truth data. The method is suitable for any linear imaging inverse problem that is ill-conditioned, and delivers results comparable to supervised conformal prediction with ground truth data.