I am a research engineer at INRAE, MIA Paris-Saclay, SOLsTIS team.

My research interests are:

  • Hidden Markov models and Bayesian inference
  • Statistical signal processing
  • Machine learning, deep learning
  • Solving problems in medical and environmental sciences

Between 2020 and 2022, I was a postdoctoral researcher at Télécom SudParis and IRISA. I worked on unsupervised statistical image segmentation approaches and unsupervised anomaly detection. Between 2017 and 2020, I was a PhD student working at Geprovas and ICube. I worked on developping new tools for the processing of images from vascular surgery. Before that, I graduated from the French engineering school Télécom SudParis in 2017.

Contact: hugo.gangloff(at)inrae(dot)fr



Code

Variational Autoencoders

  • VAE and VAE-GRF for anomaly detection [code]
  • Tutorial session on Variational Autoencoders for scRNA-seq data for DIGIT-BIO seminar [code]

Deep Hidden Markov Models

  • Unsupervised signal processing with deep extensions of Hidden Markov Chains: Python project
  • Deep parameterizations of pairwise and triplet Markov models for unsupervised classification of sequential data [code]

Some scripts and algorithm implementations

  • Experimentally show the equivalence between a gradient ascent over the EM quantity Q and a gradient ascent over the model likelihood, in the case of the training of a Hidden Markov Chain with Gaussian Independent Noise: Python script, my pdf note and see section 2.1 of this paper
  • Gaussian Markov Random Field simulation using Fourier properties and base matrices for efficiency: Python script
  • Rescaled Forward-Backward algorithm for Hidden Markov Chains with Jax using lax.scan: Python script
  • Extremely fast rescaled Forward-Backward algorithm for Hidden Markov Chains with Python calling C using ctypes and a shared library: Python script and C files
  • Chromatic Gibbs sampler for a binary Ising Markov Random Field with Jax using jit, vmap and lax.scan: Python script
  • Sequential Importance Resampling Particle Filter with Jax using jit and lax.scan: Python script


Supervising

Master 2 students



Publications

Preprints

  • Variational Autoencoder with Gaussian Random Field prior: application to unsupervised animal detection in aerial images, H. Gangloff, M.-T. Pham, L. Courtrai, S. Lefèvre, 2022, [preprint pdf]

Journal articles

  • Deep parameterizations of pairwise and triplet Markov models for unsupervised classification of sequential data, H. Gangloff, K. Morales, Y. Petetin, Computational Statistics and Data Analysis, 2022, [preprint pdf] [link] [code]
  • Unsupervised segmentation with Gaussian Pairwise Markov Fields, H. Gangloff, J.-B. Courbot, E. Monfrini, C. Collet, Computational Statistics and Data Analysis, 2021, [preprint pdf] [link]
  • Automated histological segmentation on micro-computed tomography images, S. Kuntz, H. Gangloff, H. Naamoune, A. Lejay, R. Virmani, N. Chakfé, European Journal of Vascular & Endovascular Surgery, 2021, [link]
  • Co-registration of peripheral atherosclerotic plaques assessed by conventional CT-angiography, micro-CT and histology in CLTI patients, S. Kuntz, H. Jinnouchi, M. Kutyna, S. Torii, A. Cornelissen, Y. Sato, M. E. Romero, F. Kolodgie, A. V. Finn, A. Schwein, M. Ohana, H. Gangloff, A. Lejay, N. Chakfé, R. Virmani, European Journal of Vascular & Endovascular Surgery, 2020, [link]
  • Assessing the segmentation performance of pairwise and triplet Markov models, I. Gorynin, H. Gangloff, E. Monfrini, W. Pieczynski, Signal Processing, 2017, [pdf] [link]

International conference articles

  • Variational Autoencoders for unsupervised object counting from VHR imagery: applications in dwelling extraction from IDP/refugee settlements, G.W. Gella, H. Gangloff, L. Wendt, D. Tiede, S. Lang, International Geoscience and Remote Sensing Symposium (IGARSS), 2023.
  • Weakly supervised marine animal detection from remote sensing images using vector-quantized variational autoencoder, M.T. Pham, H. Gangloff, S. Lefèvre, International Geoscience and Remote Sensing Symposium (IGARSS), 2023.
  • Object counting from aerial remote sensing images: application to wildlife and marine mammals, T. Singh, H. Gangloff, M.T. Pham, International Geoscience and Remote Sensing Symposium (IGARSS), 2023.
  • Leveraging Vector-Quantized Variational Autoencoder Inner Metrics for Anomaly Detection, H. Gangloff, M.-T. Pham, L. Courtrai, S. Lefèvre, International Conference on Pattern Recognition (ICPR), 2022, [preprint pdf]
  • A general parametrization framework for Pairwise Markov Models: an application to unsupervised image segmentation, H. Gangloff, K. Morales, Y. Petetin, International Workshop on Machine Learning for Signal Processing (MLSP), 2021, [preprint pdf], [link]
  • Unsupervised segmentation with Spatial Triplet Markov Trees, H. Gangloff, J-B Courbot, E. Monfrini, C. Collet, International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2021, [link], [slides], [poster]
  • Improved centerline tracking for new descriptors of atherosclerotic aortas, H. Gangloff, E. Monfrini, M. Z. Ghariani, M. Ohana, C. Collet, N. Chakfé, International Conference on Image Processing Theory, Tools and Applications (IPTA), 2020, [preprint pdf]
  • Unsupervised segmentation of stents corrupted by artifacts in medical X-rays images, H. Gangloff, E. Monfrini, C. Collet, N. Chakfé, International Conference on Image Processing Theory, Tools and Applications (IPTA), 2020, [preprint pdf]
  • Spatial Triplet Markov Trees for auxiliary variational inference in Spatial Bayes Networks, H. Gangloff, J-B Courbot, E. Monfrini, C. Collet, Stochastic Modeling Techniques and Data Analysis (SMTDA), 2020, [preprint pdf], [slides]
  • Performance comparison across hidden, pairwise and triplet Markov models' estimators, I. Gorynin, L. Crelier, H. Gangloff, E. Monfrini, W. Pieczynski, International Conference on Applied and Computational Mathematics (ICACM) , 2016

French conference articles

  • Chaînes de Markov cachées à bruit généralisé, H. Gangloff, K. Morales, Y. Petetin, GRETSI, 2022, [preprint pdf] (une présentation alternative des chaînes de Markov triplet généralisées)
  • Autoencodeurs variationnels à registre de vecteurs pour la détection d’anomalies, H. Gangloff, M.-T. Pham, L. Courtrai, S. Lefèvre, Reconnaissance des Formes, Image, Apprentissage et Perception (RFIAP), 2022, [pdf] (adaptation française de l'article ICPR 2022)
  • Segmentation non-supervisée dans les champs de Markov couples gaussiens, H. Gangloff, J-B Courbot, E. Monfrini, C. Collet, GRETSI, 2019
  • Segmentation de stents dans des données médicales à rayons-X corrompues par les artéfacts, H. Gangloff, E. Monfrini, C. Collet, N. Chakfe, GRETSI, 2019

See also abstract selected publications.

PhD Thesis

  • Probabilistic models for image processing: applications in vascular surgery, H. Gangloff, defended on December 15, 2020, [pdf] [slides]

Awards

  • Regional award André Blanc-Lapierre from the Electrity, Electronics and Information Scommunication Technology Society (SEE), in 2017, for the research project Stent segmentation and classification in superficial femoral artery in microCT images.
  • Second Prize for the best research internship from Mines-Télécom Foundation, in 2017, for the research project Stent segmentation and classification in superficial femoral artery in microCT images.


Teaching

See previous teaching experiences here.