Research Guillaume GINOLHAC

Home Publications Research Teaching Activities

Riemannian Geometry for Robust Estimation and Machine Learning

In this topic, the aim is to use the tools of Riemannian geometry to tackle estimation problems with constraints like subspace, structured covariance matrices, … From these geometries we are able to develop new estimation algorithms as well as new forms of center of classes (useful in machine learning). Moreover, this theory allows to obtain intrinsic Cramer Rao Bound useful to choose a correct metric and distance.

The main papers of this work are:

  • [IJ-32] F. Bouchard, A. Breloy, G. Ginolhac, A. Renaux and F. Pascal, “A Riemannian Framework for Low-Rank Structured Elliptical Models,” in IEEE Transactions on Signal Processing, vol. 69, pp. 1185-1199, 2021. https://arxiv.org/abs/2001.01141.
  • [IJ-30] F. Bouchard, A. Mian, J. Zhou, S. Said, G. Ginolhac, Y. Berthoumieu , “Riemannian geometry for Compound Gaussian distributions: application to recursive change detection”, Elsevier Signal Processing, 2020. https://arxiv.org/abs/2005.10087.
  • [IJ-25] A. Breloy, G. Ginolhac, A. Renaux, F. Bouchard, “Intrinsic Cramer-Rao Bounds for Scatter and Shape Matrices Estimation in CES Distributions”, IEEE Sig. Proc. Letters, vol. 26(2), pp 262-266, 2019. pdf project

My main collaborations in this topic are:

See also the presentation done for Finnish-French Statistical Learning for Signal Processing Mini-workshop Link on Intrinsic Bound of Scatter and Shape Matrices.

Robust Covariance and subspace Estimation

In this topic, the aim is to provide a robust estimate of the covariance. The multivariate data follows a CES distributions and we consider different a priori on the structure of the covariance matrix.

The main papers of this work are:

  • [IJ-33] A. Breloy, G. Ginolhac, Y. Gao, F. Pascal, “MIMO filters based on robust rank-constrained Kronecker covariance matrix estimation”, Elsevier Signal Processing, Vol. 187, October, 2021. pdf.
  • [IJ-21] A. Breloy, G. Ginolhac, F. Pascal et P. Forster, “Robust Covariance Matrix estimation in heterogeneous Low Rank context”, IEEE Transactions on Sig. Proc., 64 (22), pp. 5794-5806. 2016. pdf
  • [IJ-18] Y. Sun, A. Breloy, B. Prabhu, D. Palomar, F. Pascal et G. Ginolhac, “Low-Complexity Algorithms for Low Rank Clutter Parameters Estimation in Radar Systems”, IEEE Transactions on Sig. Proc., 64 (8), 2016. pdf
  • [IJ-17] A. Breloy, G. Ginolhac, F. Pascal et P. Forster, “Clutter Subspace Estimation in Low Rank Heterogeneous Noise Context”, IEEE Transactions on Sig. Proc., 63 (9), pp.2173-2182, 2015. pdf
  • [IJ-13] G. Ginolhac, P. Forster, F. Pascal et J.-P. Ovarlez, “Exploiting Persymmetry For Low-Rank Space Time Adaptive Processing”, Elsevier Sig. Proc., vol. 97, no4, pp 242 – 251, Avril 2014. pdf
  • [IJ-12] G. Ginolhac, P. Forster, F. Pascal et J.-P.Ovarlez, “Performances and Robustness of a Robust Low-Rank STAP Filter”, IEEE Transactions on Sig. Proc., Vol. 61, no1, pp 57 – 61, Janvier 2013. pdf

My main collaborations in this topic are:

For more details, see the tutorial of EUSIPCO 2018 done by Frédéric Pascal, Arnaud Breloy and Guillaume Ginolhac. Link

Robust Change Detection

In a Satellite Image Time Series, we have developed different robust detectors based on the derivation of the GLRT. The data are assumed to follow a CES distribution and the hypothesis test is built with the scatter matrices, and/or the textures. This work is partially funded by the ANR PHOENIX.

See the main papers:

  • [IJ-29] A. Mian, A. Collas, A. Breloy, G. Ginolhac, Jean-Philippe Ovarlez, “Robust low-rank change detection for SAR Image Time Series”, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, IEEE, 2020, 13, pp.3545-3556. pdf.
  • [IJ-27] A. Mian, J.P. Ovarlez, A. Atto, G. Ginolhac, “Design of New Wavelet Packets Adapted to High-Resolution SAR Images with an Application to Target Detection”, IEEE Transactions on Geoscience and Remote Sensing, 2019. pdf dataset code project
  • [IJ-26] A. Mian, G. Ginolhac, J.P. Ovarlez, A. Atto, “New Robust Statistics for Change Detection in Time Series of Multivariate SAR Images”, IEEE Transactions on Sig. Proc., vol. 67(2), pp 520-534, 2019. pdf project

My main collaborations in this topic are:

See also the tutorial done with Arnaud Breloy at IEEE RadarConf 2020 and available here (part I) and here (part II).

Robust Buried Pipe Detection by GPR

In this project, we have to detect buried pipes by a Ground Penetrating Radar. The detectors are derived from the GLRT. The data size is very large and some strategies for the covariance estimation have been proposed. This work has been funded by the FUI g4M.

See papers:

  • [IJ-22] Q. Hoarau, G. Ginolhac, A. Atto, et J.-M. Nicolas, “Robust Adaptive Detection of Buried Pipes using GPR”, Elsevier Sig. Proc., vol. 132, pp 293-305, 2017. pdf project
  • [CI-41] Q. Hoarau, A. Breloy, G. Ginolhac, A. Atto, J.-M. Nicolas et J.-P. Ovarlez, “A subspace approach for shrinkage parameter selection in undersampled configuration for Regularised Tyler Estimators”, Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. New Orleans, United States, pp.3291 – 3295, April 2017. pdf project

See presentation on this topic: link

My main collaborations in this topic are:

  • Quentin Hoarau (Past Phd Student at LISTIC – Univ. Savoie Mont-Blanc)
  • Abdourrahmane Atto (Associate professor at LISTIC – Univ. Savoie Mont-Blanc)
  • Jean Marie Nicolas (Full Professor at Telecom ParisTech)

Multilinear Algebra for RADAR Systems

In this work, we have developed a special multilinear extension of the SVD adapted to STAP and MIMO-STAP configurations. Moreover, we have derived theoretical performances of Tensorial MUSIC to show the interest of the multilinear algebra in some configuration. This work has been funded by Digiteo.

See papers:

  • [IJ-23] F. Brigui, M. Boizard, G. Ginolhac et F. Pascal, “New Low-Rank Filters for MIMO-STAP based on an Orthogonal Tensorial Decomposition”, IEEE Transactions on Aerospace and Electronic System, vol. 54(3), pp 1208-1220, 2018. pdf
  • [IJ-20] P. Forster, G. Ginolhac, M. Boizard, “Derivation of the theoretical performance of a Tensor MUSIC Algorithm”, Elsevier Sig. Proc., vol. 129, no12, pp 97-105, december 2016. pdf
  • [IJ-16] M. Boizard, G. Ginolhac, F. Pascal et P. Forster, “Low-Rank Filter and Detector for Multidimensional Data Based on an Alternative Unfolding HOSVD, Application to Polarimetric STAP”, EURASIP Journal on Advances in Signal Processing, SpringerOpen, pp.119-130, 2014.  pdf

My main collaborations in this topic are:

  • Frédéric Pascal (Full professor at LSS – CentraleSupelec)
  • Frédéric Brigui (Researcher at ONERA)
  • Philippe Forster (Full professor at SATIE – ENS Cachan)
  • Maxime Boizard (Past PhD student at SATIE – ENS Cachan)

Random Matrix Theory Tools for RADAR algorithms

In this topic, we have used the Random Matrix Theory tools to derive theoretical approximations of performance of filters and detectors when the disturbance is composed of strong low-rank contribution. Next, we want to propose new low-rank detectors which keep good asymptotic properties in a large dimension regime.

See papers:

  • [CI-47] P. Vallet, G. Ginolhac, F. Pascal, P. Forster, « An Improved Low Rank Detector in the High Dimensional Regime », ICASSP 2019, Brighton, England. pdf project
  • [IJ-28] [IJ-28] A. Combernoux, F. Pascal, G. Ginolhac, M. Lesturgie “Theoretical Performance of Low Rank Adaptive Filters in the Large Dimensional Regime”, IEEE Transactions on Aerospace and Electronic Systems, vol. 55(6), pp. 1557-9603, December 2019. pdf
  • [IJ-24] A. Combernoux, F. Pascal, G. Ginolhac, “Performance of the Low-Rank Adaptive Normalized Matched Filter Test Under a Large Dimension Regime”, IEEE Transactions on Aerospace and Electronic Systems, vol. 55(1), pp, 459-468, 2019. pdf

My main collaborations in this topic are:

  • Frédéric Pascal (Full professor at LSS – CentraleSupelec)
  • Pascal Vallet (Associate professor at IMS – Univ. de Bordeaux)
  • Alice Combernoux  (Past PhD student at ONERA)
  • Rémi Beisson (Current PhD student at Univ. of Bordeaux)

SAR processors based on Matched Subspace Detectors

In this work, we have developed new SAR detectors based on Matched Subspace Detectors. In particular Man Made Targets (MMT) are assumed to lie in a low rank susbpace as well as the interference (forest).

See papers:

  • [IJ-15] F. Brigui, G. Ginolhac, L. Thiron-Lefèvre et P. Forster, “New SAR Algorithm based on Signal and Orthogonal Interference Subspace Models”, IEEE Transactions on Geoscience and Remote Sensing, Vol. 52, no7, pp 3800 – 3811, July 2014. pdf
  • [IJ-14] F. Brigui, G. Ginolhac, L. Thiron-Lefevre et P. Forster, “New SAR Algorithm based on Oblique Projection for Interference Reduction”, IEEE Transactions on Aerospace and Electronic System, Vol. 50, no2, pp.1118 – 1137, Avril 2014. pdf
  • [IJ-9] R. Durand, G. Ginolhac, L. Thirion-Lefèvre et P. Forster, “Back Projection Version of Subspace Detector SAR Processors”, IEEE Trans. on Aero. Space and Elec. Syst., vol. 47, no2, Avril 2011. pdf
  • [IJ-8] G. Ginolhac, L. Thirion-Lefèvre, R. Durand et P. Forster, “SAR Processors Based on Signal or Interference Subspace Detector”, IEEE Trans. on Aero. Space and Elec. Syst., vol. 46, no3, pp 1006-1020, Juillet 2010. pdf
  • [IJ-4] R. Durand, G. Ginolhac, L. Thirion et P. Forster, “New SAR Processor Based on Matched Subspace Detectors”, IEEE Trans. on Aero. Space and Elec. Syst., vol. 45, no1, pp 221-236, Janvier 2009. pdf

My main collaborations in this topic are:

  • Laetitia Thirion Lefevre (Full professor at SONDRA – CentraleSupelec)
  • Frédéric Brigui (Past PhD student at SONDRA – CentraleSupelec)
  • Philippe Forster (Full professor at SATIE – ENS Cachan)
  • Rémi Durand (Past PhD student at SONDRA – CentraleSupelec)

____________________________________________________

Current PhD students :

Olivier Lerda, Robust Statistics for Underwater SONAR

  • Doctorant Ecole Doctorale SISEO from University Savoie Mont-Blanc, started in 2021 March 1st
  • Other supervisor : Ammar Mian (Univ. Savoie Mont-Blanc) and Jean Philippe Ovarlez (ONERA)

Rémi Beisson, Change detection and classification in high dimensional multivariate image times series

  • Doctor from University of Bordeaux, started in 2020 October 1st
  • Other supervisors: Pascal Vallet and Audrey Giremus from IMS (University of Bordeaux)

Hoa VU PHAN VIET, Statistical learning for Earth deformation monitoring with satellite InSAR

  • Doctorant Ecole Doctorale SISEO from University Savoie Mont-Blanc, started in 2020 October 1st
  • Other supervisors: Yajing Yan from LISTIC, Arnaud Breloy from LEME and Frédéric Brigui from ONERA

Antoine Collas, Clustering from SAR time series images

  • Doctorant Ecole Doctorale STIC from University Paris-Saclay, started in 2019 October 1st
  • Other supervisors: Jean-Philippe Ovarlez and Chengfang Ren from SONDRA and Arnaud Breloy from LEME
  • Publications : [IJ-29], [CI-55]

Past PhD students :

Ammar Mian, (SONDRA funding), “Contributions to SAR Image Time Series Analysis “

  • Doctorant Ecole Doctorale STIC from University Paris-Saclay, defended september 26th 2019 (started in 2016 October 1st)
  • Other supervisors: Jean-Philippe Ovarlez from SONDRA and Abdou Atto from LISTIC
  • Composition of the Jury :
    • Frédéric Pascal, Professeur at CentraleSupelec, France (Président, section 61)
    • Jean Yves Tourneret, Professeur Univ. de Toulouse (Rapporteur, section 61)
    • André Ferrari, Professeur Univ. de Nice (Rapporteur, section 61)
    • Maria S. Greco, Professor, Univ. de Pisa, Italy, (Examinateur, section 61)
    • Jean Philippe Ovarlez, Maître de Recherche ONERA / SONDRA (Directeur de thèse, section 61)
    • Guillaume Ginolhac, PR University Savoie Mont-Blanc, Annecy (Co-directeur, section 61)
    • Abdourahmane Atto (co-encadrant, section 61)
  • Publications : [IJ-26], [IJ-27], [CI-50], [CI-48], [CI-45], [CI-44], [CI-43], [CN-18], [CN-20]
  • Situation en 2021 : Associate Professor at University Savoie Mont-Blanc.

Quentin Hoarau, (FUI funding), “Robust adaptive Detection of buried pipes by means of a GPR radar”

  • Doctorant Ecole Doctorale SISEO from University Savoie Mont-Blanc, started in 2014 November 1st
  • Other supervisors: Abdou Atto from LISTIC and Jean-Marie Nicolas from Télécom ParisTech
  • Publications : [IJ-22], [CI-41], [CI-39], [CN-17]
  • Q. Hoarau gave up his PhD during its 5th year.

Alice Combernoux (SONDRA funding), “Low Rank filtering and detection using random matrix theory tools.”

  • Doctorant Ecole Doctorale STITS from University Paris XI, defended january 29th 2016 (started october 1st 2012)
  • Composition of the Jury :
    • Sylvie Marcos, DR CNRS LSS, CentraleSupelec, (President, section 61)
    • Walid Hachem, DR CNR LTCI, TelecomsParisTech (Rapporteur, section 61)
    • François Vincent, Professeur at ISAE, Toulouse (Rapporteur, section 61)
    • Chong Chin Yuan, chercheur au DSO, Singapore, (Examinateur, section 61)
    • Romain Couillet, associate professor, CentraleSupelec, (Examinateur, section 61)
    • François Le Chevalier, Professeur at Delft University, Netherlands, (Examinateur, section 61)
    • Guillaume Ginolhac, PR University Savoie Mont-Blanc, Annecy (Co-supervisor, section 61)
    • Frederic Pascal, Professeur at CentraleSupelec, France (Co-supervisor, section 61)
    • Marc Lesturgie, Chercheur SONDRA, CentraleSupelec (Supervisor, section 61)
  • Publications : [IJ-24], [IJ-28], [CI-36], [CI-34], [CI-30], [CN-14], [S-12]
  • Situation en 2016 : Engineer at Thales Underwater Systems, Brest, FRANCE.


Arnaud Breloy (DGA funding), “Algorithmes of estimation and detection in a Low Rank heterogenous context.”

  • Doctorant Ecole Doctorale Sciences Pratiques (EDSP) of ENS Cachan, defended november 23th 2015 (started october 1st 2012)
  • Composition du Jury :
    • Pierre Comon, DR CNRS at Gipsa Lab, Grenoble (President, section 61)
    • Olivier Besson, Professeur at ISAE, Toulouse (Rapporteur, section 61)
    • Pascal Chevalier, Professeur au CNAM et chercheur at THALES (Rapporteur, section 61)
    • Chong Chin Yuan, chercheur au DSO, Singapore, (Examinateur, section 61)
    • Guillaume Ginolhac, PR University Savoie Mont-Blanc, Annecy (Supervisor, section 61)
    • Frederic Pascal, Professeur at CentraleSupelec, France (Co-supervisor, section 61)
    • Philippe Forster, PR de l’Univ. Paris Ouest La Défense, Ville d’Avray (Co-supervisor, section 61)
  • Publications avec le doctorant : [IJ-21], [IJ-18], [IJ-17], [CI-38], [CI-35], [CI-33], [CI-32], [CI-29],
    [CN-13]
  • Situation en 2016 : Associate professor at University Paris Ouest La Défense, FRANCE.

Maxime Boizard (digiteo funding), “Development and performance studies of detectors in multidimensional RADAR configurations”

  • Doctorant Ecole Doctorale Sciences Pratiques (EDSP) de l’ENS Cachan, defended december 13th 2015 (started october 1st 2010)
  • Composition du Jury :
    • David Brie, PR University of Lorraine, Nancy (President, section 61)
    • Pierre Comon, DR CNRS au Gipsa Lab, Grenoble (Rapporteur, section 61)
    • Sylvie Marcos, DR CNRS au LSS, Palaiseau (Rapporteur, section 61)
    • Pascal Chevalier, Professeur au CNAM et chercheur at THALES (Examinateur, section 61)
    • Guillaume Ginolhac, PR University Savoie Mont-Blanc, Annecy (Co-supervisor, section 61)
    • Frederic Pascal, Professeur at CentraleSupelec, France (Co-supervisor, section 61)
    • Philippe Forster, PR de l’Univ. Paris Ouest La Défense, Ville d’Avray (Supervisor, section 61)
  • Publications avec le doctorant : [RI-23], [RI-20], [RI-16], [CI-31], [CI-28], [CI-27], [CI-25], [CN-11],[CN-12]
  • Situation in 2016 : Engineer at Akka, Toulouse, FRANCE.

Frédéric Brigui (DGA funding), “Algorithmes of Polarimetric SAR Imagery  based on subspace models: Application to FOPEN detection”

  • Doctorant “Connaissance, langage, modélisation” de l’Univ. Paris Ouest La Défense, defended december 3rd 2010 (started october 1st 2007)
  • Composition du Jury :
    • Jocelyn Chanussot, PU at GIPSA-Lab, Grenoble (President, section 61)
    • François Vincent, Enseignant Chercheur at ISAE, Toulouse (Rapporteur, section 61)
    • Ridha Touzi, Chercheur au Centre Recherche de Télédection, Canada (Rapporteur, section 63)
    • Marc Lesturgie, Directeur de SONDRA, SUPELEC, France (Examinateur, section 63)
    • Carole Nahum, Engineer DGA (Examinateur)
    • Guillaume Ginolhac, MCF IUT Ville d’Avray, Ville d’Avray (Co-supervisor, section 61)
    • Laetitia Thiron-Lefèvre, Chercheur at SONDRA, SUPELEC, France (Co-supervisor, section 63)
    • Philippe Forster, PU de l’Univ. Paris Ouest La Défense, Ville d’Avray (Supervisor, section 61)
  • Publications : [RI-15], [RI-14], [RI-7], [RN-2], [CI-23], [CI-22], [CI-21], [CI-20], [CI-18], [CI-17], [CN10], [CN-9], [CN-8], [W-1]
  • Situation in 2016 : Resarcher at ONERA, Palaiseau.

Rémi Durand (SONDRA funding), SAR processors based on subspace detectors

  • Doctorant “Connaissance, langage, modélisation” de l’Univ. Paris Ouest La Défense, defended november 30th 2007 (started october 1st 2004) obtenaid with the felicitations of the jury
  • Composition du Jury :
    • Olivier Besson, Chercheur at ISAE, Toulouse (President et Rapporteur, section 61)
    • Florence Tupin, Enseignant Chercheur at ENST Paris (Rapporteur, section 63)
    • François Le Chevalier, Engineer at THALES (Examinateur, section 61)
    • Marc Lesturgie, Directeur of SONDRA, SUPELEC, France (Examinateur, section 63)
    • Guillaume Ginolhac, MCF IUT Ville d’Avray, Ville d’Avray (Co-supervisor, section 61)
    • Laetitia Thiron-Lefèvre, Chercheur at SONDRA, SUPELEC, France (Co-supervisor, section 63)
    • Philippe Forster, PU de l’Univ. Paris Ouest La Défense, Ville d’Avray (Supervisor, section 61)
  • Publications : [RI-9], [RI-8], [RI-4], [RN-2], [CI-13], [CI-12], [CI-11], [CI-10], [CI-9] et [CN-4]
  • Situation in 2012 : Engineer at THALES Underwater Systems, Nice, France.