Charlesquin Kemajou Mbakam

Make the change you want to see

In 2015, I obtained a master’s degree in Applied Mathemitics from the Universy of Dschang where my work focused on modelling, analyzing and smulating life cycle of Pultella Xylostella. In 2020, I completed another master’s degree in Artificial Intelligence at the African Masters of Machine intelligence (AMMI) in Rwanda, working on semi-blind inverse problems in imaging. Following this, I pursued a PhD in Statistics and machine learning at Heriot-Watt University in Edinburgh, under the supervision of Professors Marcelo Pereyra and Jean-François Giovannelli. During my doctoral studies, my work focused on developing advanced methodologies integrating statistical modelling with deep learning in a unified framework to address inverse problems in imaging.

Currently, my research centers on Bayesian inverse problems, with a particular emphasis on leveraging data-driven priors to tackle complex challenges in computational imaging. My work bridges advanced statistical techniques and machine learning approaches, aiming to enhance the accuracy and efficiency of image reconstruction and analysis.

news

Dec 14, 2025	I am deligthed to informed you that I will be starting a new role as Research Associate at heriot-Watt University in January 2026. This exciting opportunity will allows me delve deeply into developing reliable software for image inverse problems, with application spanning medicine, agriculture and astronomic.
Sep 17, 2024	I successfully passed my PhD viva, with Prof. Audrey Repetti as the internal examiner and Prof. Martin Benning as the external examiner.

latest posts

Feb 17, 2025	Common statistical ways to measure distance between two distributions

selected publications

SIAM

Marginal Likelihood Estimation in Semiblind Image Deconvolution: A Stochastic Approximation Approach

Charlesquin Kemajou Mbakam, Marcelo Pereyra, and Jean-François Giovannelli

SIAM Journal on Imaging Sciences, 2024

Bib HTML Code

@article{mbakam2024marginal,
  title = {Marginal Likelihood Estimation in Semiblind Image Deconvolution: A Stochastic Approximation Approach},
  author = {Mbakam, Charlesquin Kemajou and Pereyra, Marcelo and Giovannelli, Jean-Fran{\c{c}}ois},
  journal = {SIAM Journal on Imaging Sciences},
  volume = {17},
  number = {2},
  pages = {1206--1254},
  year = {2024},
  publisher = {SIAM},
}

JMIV

Empirical Bayesian image restoration by Langevin sampling with a denoising diffusion implicit prior

Charlesquin Kemajou Mbakam, Jean-Francois Giovannelli, and Marcelo Pereyra

arXiv preprint arXiv:2409.04384, 2024

Bib HTML Code

@article{mbakam2024empirical,
  title = {Empirical Bayesian image restoration by Langevin sampling with a denoising diffusion implicit prior},
  author = {Mbakam, Charlesquin Kemajou and Giovannelli, Jean-Francois and Pereyra, Marcelo},
  journal = {arXiv preprint arXiv:2409.04384},
  year = {2024},
}

ARXIV

Do Bayesian imaging methods report trustworthy probabilities?

David YW Thong, Charlesquin Kemajou Mbakam, and Marcelo Pereyra

arXiv preprint arXiv:2405.08179, 2024

Bib HTML Code

@article{thong2024bayesian,
  title = {Do Bayesian imaging methods report trustworthy probabilities?},
  author = {Thong, David YW and Mbakam, Charlesquin Kemajou and Pereyra, Marcelo},
  journal = {arXiv preprint arXiv:2405.08179},
  year = {2024},
}

TMLR

Learning few-step posterior samplers by unfolding and distillation of diffusion models

Charlesquin Kemajou Mbakam, Marcelo Pereyra, and Jonathan Spence

Transactions on Machine Learning Research, 2025

Bib HTML Code

@article{mbakam2025learning,
  title = {Learning few-step posterior samplers by unfolding and distillation of diffusion models},
  author = {Mbakam, Charlesquin Kemajou and Pereyra, Marcelo and Spence, Jonathan},
  journal = {Transactions on Machine Learning Research},
  year = {2025},
  pages = {2835-8856},
}