pageHPM-2017

Hélène Paugam-Moisy Computer Science

Professor, Université des Antilles
Head of AID research team at LAMIA
Pointe-à-Pitre, Guadeloupe, France

Previously:
Professor at Université Lyon 2 -- Lyon, France
Research position at INRIA Saclay - Île-de-France
TAO team, at Université Paris-Sud -- Orsay, France

Research -- CV -- Funding -- Management -- Teaching -- Publications -- Contact

My main scientific interest is understanding how knowledge is acquired, represented, stored, processed and retrieved in the brain, through computational modelling and learning in artificial neural networks.

Competence and skill in several research areas: Machine Learning -- Spiking Neuron Networks -- Parallel Computing -- Cognitive Science

Research

My current topics of exploration are Spiking Neuron Networks, especially Polychronization and Reservoir Computing, and Machine Learning, especially Deep Neural Networks.

Spiking Neuron Networks are the latest generation of neural networks. They are strongly inspired by neuronal processing in the brain. Taking into account the precise spike-timing of neurons, these models have the capacity to process spatio-temporal data.
Reservoir Computing is a family of recurrent neural networks that let freely the dynamics of an internal network (the "reservoir") compute an expansion function of temporal input data, with fading memory. Internal states can be compared to echos or ripples on water. Although learning only the readout weights can be sufficient, I investigate the effect of synaptic plasticity in a reservoir of spiking neurons, regarding it as a complex system.
Deep Neural Networks are multilayer feedforward neural networks, revisited with new layerwise learning rules. I study how to optimize their architecture and learning performance. Mainly I investigate the sparsity in processing internal features.

CV and previous activities

With a background in Mathematics (agregation, 1987), I have first been a teacher in french education. I obtained a PhD in Computer Science (1992) and an HDR (1997) at Ecole Normale Supérieure de Lyon. When I started research at the LIP-CNRS laboratory (ENS Lyon, 1989-1998), I was impressed by the huge capacity of parallel processing, both in computers and in the brain. Furthermore I was captivated by the notions of learning and memory in artificial systems and I became really interested in Machine Learning. At LIP, I was head of the research group Connectionism and Machine Learning. I started to learn about Cognitive Science and I joined Université Lyon 2 as a full professor in 1998. I was head of the research team Connectionism and Cognitive Modelling at the CNRS Institute for Cognitive Science (1998-2006). Recently I was a Professor at Université Lyon 2, member of LIRIS-CNRS and I joined the Thème Apprentissage et Optimisation (TAO) in 2008, for a research position at INRIA Saclay - Île-de-France. I joined Université des Antilles in the mid 2016.

Contracts and funded projects

ANR-DEFIS, ASAP: "Apprentissage Statistique par une Architecture Profonde / Statistical Learning in Deep Networks" (2009-2012), scientific head of the TAO-INRIA partner.
ANR-Blanc, MAPS: "Mappings, Adaptation, Plasticity and Spatial computation" (2007-2010), scientific head of the LIRIS partner.
ACI-NIC, "Distributed models of spiking neuron networks for multimodal integration” + DYNN (2003-2006)
ACI-Cognitique, "Action and causality" (2001-2003) + "Spatial cognition" (1999-2001)
European Network of Excellence NEUROCOLT (1 & 2) (1994-2002)
Contract with ELF-TOTAL (1997 and 2000)
European Project PARALIN, INCO-DC (1995-1998)
Contract with France Télécom (1994-1997)

Main responsibilities and research animation

Independent expert for the European Commission, FP6 (2003 and 2005) and FP7 (2007 and 2012).
Member of the "Comité National du CNRS", section 07 (Computer Science, Signal Processing, Automatics) and CID 45 (Cognition), 2004-2008
Co-head of the PRASC: animation of research and project funding in Cognitive Science, in Rhône-Alpes, 2000-2007
Head of the Cognitive Science Master of Lyon (2000-2004), co-head of the Computer Science Master of Lyon (1993-1998)
Organization of Special Sessions at ESANN, European Symposium in Artificial Neural Networks: "Neural Netwoks and Cognitive Science" (2002), "Neural Netwoks for Computational Neuroscience" (2008), "Deep Learning" (2011)
Organizing committee of ECAI'2002, European Conference in Artificial Intelligence
Member of international conferences program committees (e.g. NIPS, IJCNN, ICANN, ESANN, IJCAI, ICML, ACML)

PhD advisor

Emmanuel BIABIANY, started in 2016
Stéphane CHOLET, started in 2015

Recently:

Ludovic ARNOLD, PhD in 2013, Université Paris-Sud
Régis MARTINEZ, PhD in 2011, Université Lyon 2
Sylvain CHEVALLIER, PhD in 2009, Université Paris-Sud
Anthony MOURAUD, PhD in 2009, Université des Antilles
David MEUNIER, PhD in 2007, Université Lyon 2
Johnny MARIETHOZ, PhD in 2006, Université Lyon 2
Hyungsung JI, PhD in 2004, Université Lyon 2
Emanuelle REYNAUD, PhD in 2002, Université Lyon 2
Thomas VOEGTLIN, PhD in 2002, Université Lyon 2
Olivier TEYTAUD, PhD in 2001, ENS Lyon
André ELISSEEFF, PhD in 2000, ENS Lyon

Main courses

Master module on Neural Networks and Machine Learning
Master module on Artificial Intelligence
Master module on Distributed and Parallel Computing
Master module on Neural Networks and Associative Memories
Master module on Statistical Learning Theory
Course on Mathematics for Computer Science
Course on Modelling and Cognition
Course on Networks and Distributed Systems
Course on Parallel Programming Languages (PVM, MPI)
Courses on programming languages (Pascal, C, C++, Java)

List of selected publications

Book chapters:

S. Rebecchi, H. Paugam-Moisy and M. Sebag. Learning sparse features with an auto-associator. In T. Kowaliw, R. Doursat and N. Bredeche, Eds, Growing Adaptive Machines, 139–158, Springer, 2014.

H. Paugam-Moisy and S.M. Bohte. Computing with Spiking Neuron Networks. In G. Rozenberg and T.H.W. Bäck and J.N. Kok, Eds, Handbook of Natural Computing, 335–376. Springer, 2012.

Y. Guermeur et H. Paugam-Moisy. Théorie de l’apprentissage de Vapnik et SVM, Support Vector Machines. Dans M. Sebban et G. Venturini, éditeurs, Apprentissage automatique, pages 109–138. Hermès: Paris, 1999.

H. Paugam-Moisy. Multiprocessor simulation of neural networks. In M.A. Arbib, editor, The Handbook of Brain Theory and Neural Networks, pages 605–608. The MIT Press, 1995.

H. Paugam-Moisy. Parallel neural computing based on network duplicating. In I. Pitas, editor, Parallel Algorithms for Digital Image Processing, Computer Vision and Neural Networks, pages 305–340. John Wiley, 1993.

H. Paugam-Moisy and A. Pétrowski. Parallel neural computing based on algebraic partitioning. In I. Pitas, editor, Parallel Algorithms for Digital Image Processing, Computer Vision and Neural Networks, pages 259–304. John Wiley, 1993.

Journal articles:

B. Cessac, H. Paugam-Moisy and T. Viéville. Overview of facts and issues about neural coding by spikes. Journal of Physiology: Paris. (35 pages - to appear). 2009

H. Paugam-Moisy, R. Martinez and S. Bengio. Delay learning and polychronization for reservoir computing. NeuroComputing, 71: 1143-1158, Elsevier, 2008.

Y. Guermeur, G. Pollastri, A. Elisseeff, D. Zelus, H. Paugam-Moisy and P. Baldi. Combining Protein Secondary Structure Prediction Models with Ensemble Methods of Optimal Complexity. NeuroComputing, 56: 305–327, Elsevier, 2004.

J.-P. Royet, O. Koenig, H. Paugam-Moisy, D. Puzenat and J.-L. Chasse. Levels of processing effects on a task of olfactory naming. Perceptual & Motor Skills, 98: 197-213, 2004.

P. Estévez, H. Paugam-Moisy, D. Puzenat and M. Ugarte. A scalable parallel algorithm for training a hierarchical mixture of experts. Parallel Computing, 28: 861-891. Elsevier, 2002.

E. Reynaud, A. Crépet, H. Paugam-Moisy and D. Puzenat. A computational model for binding sensory modalities. Abstract in Consciousness and Cognition, 9(2): 87–88. Academic Press, 2000.

A. Elisseeff and H. Paugam-Moisy. JNN, a Randomized Algorithm for Training Multilayer Networks in Polynomial Time. NeuroComputing, 29: 3–24, Elsevier, 1999.

C. Kenyon and H. Paugam-Moisy. Multilayer neural networks and polyhedral dichotomies. Annals of Mathematics and Artiﬁcial Intelligence, 24(1-4): 115–128, Baltzer, 1998.

J.-P. Royet, H. Paugam-Moisy, C. Rouby, A.D. Zighed, N. Nicoloyannis, S. Amghar, and G. Sicard. Is short-term odour recognition predictable from odour proﬁle ? Chemical Senses, 21 :553–566, Oxford University Press, 1996.

M. Cosnard, P. Koiran, and H. Paugam-Moisy. Bounds on the number of units for computing arbitrary dichotomies by multilayer perceptrons. Journal of Complexity, 10: 57–63, Academic Press, 1994.

International conference papers:

Mouraud, A. and Barthelemy, Q. and Mayoue, A. and Gouy-Pailler, C. and Larue, A. and Paugam-Moisy, H. From neuronal cost-based metrics towards sparse coded signal classification. In Proc of ESANN’2012, M. Verleysen Ed., 311–316. Bruges, Belgium, April 2012.

Chevallier, S. and Bredeche, N. and Paugam-Moisy, H. and Sebag, M. Emergence of Temporal and Spatial Synchronous Behaviors in a Foraging Swarm. In Proc. of European Conference in Artificial Life, ECAL’2011, 125–132, MIT Press. Paris, France. August 2011.

L. Arnold and S. Rebecchi and S. Chevallier and H. Paugam-Moisy. An Introduction to Deep Learning [tutorial of special session]. In Proc. of ESANN’2011, Advances in Computational Intelligence and Learning, M. Verleysen Ed., 477–488. Bruges, Belgium, April 2011.

Chevallier, S. and Paugam-Moisy, H. and Sebag, M. SpikeAnts, a spiking neuron network modelling the emergence of organization in a complex system. In Advances in Neural Information Processing Systems 23, NIPS’2010, 379–387. Vancouver, Canada. December 2010.

Mouraud, A. and Guillaume, A. and Paugam-Moisy, H. The DAMNED simulator for implementing a dynamic model of the network controlling saccadic eye movements. In Artificial Neural Networks, ICANN’2010, Lecture Notes in Computer Science 6352 :272-281, Springer. Thessaloniki, Greece. September 2010.

Arnold, L. Paugam-Moisy, H. and Sebag, M. Unsupervised Layer-Wise Model Selection in Deep Neural Networks. In European Conference on Artificial Intelligence, ECAI’2010, Frontiers in Artificial Intelligence and Applications 215 :915-920, IOS Press Amsterdam. Lisbon, Portugal. August 2010.

R. Martinez and H. Paugam-Moisy. Algorithms for structural and dynamical polychronous groups detection. In Proc. of ICANN'2009, Lecture Notes in Computer Science 5769, Springer, 75-84. Limassol, Cyprus, Sepember 2009.

D. Meunier and H. Paugam-Moisy. Neural networks for computational neuroscience [tutorial]. In Proc. of ESANN’2008, Advances in Computational Intelligence and Learning, M. Verleysen Ed., 367–378. Bruges, Belgium, April 2008.

H. Paugam-Moisy, R. Martinez and S. Bengio. A supervised learning approach based on STDP and polychronization in spiking neuron networks. In Proc. of ESANN’2007, Advances in Computational Intelligence and Learning, M. Verleysen Ed., 427–432. Bruges, Belgium, April 2007.

A. Mouraud, H. Paugam-Moisy and D. Puzenat. DAMNED: A distributed and multithreaded neural event driven simulation framework. In Proc. of Int. Conf. on Parallel Distributed Computing and Networks, PDCN’2006, ACTA Press, 212–217. Innsbruck, Austria, February 2006.

D. Meunier and H. Paugam-Moisy. Cluster detection algorithm in neural networks. In Proc. of Europ. Symp. on Artiﬁcial Neural Networks, ESANN’2006, M. Verleysen Ed., 19–24. Bruges, Belgium, April 2006.

S. Chevallier, P. Tarroux and H. Paugam-Moisy. Saliency extraction with a distributed spiking neuron network. In Proc. of Europ. Symp. on Artiﬁcial Neural Networks, ESANN’2006, M. Verleysen Ed., 209–214. Bruges, Belgium, April 2006.

A. Mouraud and H. Paugam-Moisy. Learning and discrimination through STDP in a top-down modulated associative memory. In Proc. of Europ. Symp. on Artiﬁcial Neural Networks, ESANN’2006, M. Verleysen Ed., 611–616. Bruges, Belgium, April 2006.

D. Meunier and H. Paugam-Moisy. Evolutionary supervision of a dynamical neural network allows learning with on-going weights. In Proc. of Int. Joint Conf. on Neural Networks, IJCNN’2005, IEEE-INNS, 1493–1498. Montréal, Canada, July 2005.

E. Reynaud and H. Paugam-Moisy. A multiple BAM for hetero-association and multisensory integration modelling. In Proc. of Int. Joint Conf. on Neural Networks, IJCNN’2005, IEEE-INNS, 2117–2122. Montréal, Canada, July 2005.

S. Chevallier, H. Paugam-Moisy and F. Lemaître. Distributed processing for modelling real-time multimodal perception in a virtual robot. In Proc. of Int. Conf. on Parallel Distributed Computing and Networks, PDCN’2005, ACTA Press, 393–398. Innsbruck, Austria, February 2005.

D. Meunier and H. Paugam-Moisy. A "spiking" Bidirectional Associative Memory for modeling intermodal priming. In Proc. of Int. Conf. on Neural Networks and Computational Intelligence, NCI’2004, ACTA Press. 25–30. Grindelwald, Switzerland, February 2004.

Bouchut, Y., Paugam-Moisy, H. and Puzenat, D. Asynchrony in a distributed modular neural network for multimodal integration. In Proc. of Int. Conf. on Parallel and Distributed Computing & Systems, PDCS’2003, ACTA Press, 588-593. Marina del Rey, USA, November 2003.

H. Paugam-Moisy, D. Puzenat, E. Reynaud and J.-Ph. Magué. Neural networks for modelling memory : Case studies [tutorial]. In Proc. of European Conference on Artiﬁcial Neural Networks, ESANN’2002, D-facto, 71-82. Bruges, Belgium, April 2002.

O. Teytaud and H. Paugam-Moisy. Bounds on the generalization ability of Bayesian inference and Gibbs algorithms. In Proc. of International Conference on Artiﬁcial Neural Networks, ICANN’2001, Lecture Notes in Computer Science 2130, Springer, 263–268. Vienne, Austria, August 2001.

H. Paugam-Moisy and E. Reynaud. Multi-network system for sensory integration. In Proc. of Int. Joint Conf. on Neural Networks, IJCNN’2001, IEEE-INNS, 2343–2348. Washington DC, US, July 2001.

Y. Guermeur, A. Elisseeff and H. Paugam-Moisy. A new multi-class SVM based on a uniform convergence result. In Proc. of International Joint Conference on Neural Networks, IJCNN’2000, IEEE-INNS, IV :183–188. Como, Italy, July 2000.

H. Paugam-Moisy, A. Elisseeff and Y. Guermeur. Generalization performance of multiclass discriminant models. In Proc. of IJCNN’2000, IEEE-INNS, IV :177-182. Como, Italy, July 2000.

A. Crépet, H. Paugam-Moisy, E. Reynaud and D. Puzenat. A modular neural network for binding several modalities. In Proc. of International Conference on Artiﬁcial Intelligence, IC-AI’2000, 921–928. CSREA Press. Las Vegas, US, June 2000.

E. Reynaud, A. Crépet, H. Paugam-Moisy and D. Puzenat. A modular neural network model for a multi-modal associative memory. In Proc. of 4th International Conference on Cognitive and Neural Systems, ICCNS’2000, 17. Boston, US, June 2000.

Y. Guermeur, A. Elisseeff, and H. Paugam-Moisy. Estimating the sample complexity of a multiclass discriminant model. In Proc. of 9th International Conference on Artiﬁcial Neural Networks, ICANN’99, 310–315. IEE, 1999.

C. Bertolini, H. Paugam-Moisy, and D. Puzenat. Priming an Artiﬁcial Associative Memory. In J. Mira and J.V. Sanchez-Andres, editors, Proc. of International Workshop on Artiﬁcial Neural Networks, IWANN’99, LNCS, 1606 :348–356. Springer, 1999.

Y. Guermeur, H. Paugam-Moisy, and P. Gallinari. Multivariate Linear Regression on Classiﬁer Outputs : A Capacity Study. In L. Niklasson, L. Bodén and T. Ziemke, editors, Proc. of 8th International Conference on Artiﬁcial Neural Networks, ICANN’98, 693–698. Springer, 1998.

A. Elisseeff, and H. Paugam-Moisy. Jacobian Neural Network Learning Algorithms. In L. Niklasson, L. Bodén and T. Ziemke, editors, Proc. of 8th International Conference on Artiﬁcial Neural Networks, ICANN’98, 573–578. Springer, 1998.

H. Paugam-Moisy. Neural networks : from massively parallel processing towards modular distributed processing. Invited talk, in Proc. of VIII Congresso Latinoamericano de Control Automatica, 1: 31–36. Viña del Mar, CHILI. 1998.

P. Estevez, M. Ugarte, H. Paugam-Moisy, and D. Puzenat. Modular Parallel Implementation for HME Neural Networks. In Proc. of 1998 Int. Conf. on Parallel and Distributed Processing Technique and Applications, PDPTA’98, III :1365–1372. CSREA Press, ISBN 1-892512-06-8, 1998.

P. Estevez, M. Ugarte, H. Paugam-Moisy, and D. Puzenat. Parallel Simulation of Modular Neural Networks. In Proc. of XII Congreso Chileno de Ingenieria Electrica, 555–560. Universidad de La Frontera, Temuco, CHILI, 1997.

G. Brightwell, C. Kenyon, and H. Paugam-Moisy. Multilayer neural networks : one or two hidden layers ? In M.C. Mozer, M.I. Jordan, and T. Petsche, editors, Advances in Neural Information Processing Systems 9, Proc. NIPS*96, 148–154. MIT Press, 1997.

A. Elisseeff and H. Paugam-Moisy. Size of multilayer networks for exact learning : analytic approach. In M.C. Mozer, M.I. Jordan, and T. Petsche, editors, Advances in Neural Information Processing Systems 9, Proc. NIPS*96, 162–168. MIT Press, 1997.

V. Demian, F. Desprez, H. Paugam-Moisy, and M. Pourzandi. Parallel implementation of RBF neural networks. In Parallel Processing, Proc. EuroPar’96, LNCS, 1124 :243–250. Springer, 1996.

B. Girau and H. Paugam-Moisy. Load sharing in the training set partition algorithm for parallel neural learning. In 9th Int. Parallel Processing Symposium, Proc. IPPS’95, 586–591. IEEE Press, 1995.

D. Girard and H. Paugam-Moisy. Strategies of weight updating for parallel back-propagation. In C. Girault, editor, Applications in Parallel and Distributed Computing, volume A-44 of IFIP Transactions, 335–336. North-Holland, 1994.

A. Azcarraga, H. Paugam-Moisy, and D. Puzenat. An incremental neural classiﬁer on a MIMD parallel computer. In C. Girault, editor, Applications in Parallel and Distributed Computing, volume A-44 of IFIP Transactions, 13–22. North-Holland, 1994.

M. Cosnard, P. Koiran, and H. Paugam-Moisy. A step towards the frontier between one-hidden layer and two-hidden layer neural networks. In Proc. Int. Joint Conf. Neural Networks, IJCNN’93, 3 :2292–2295. Nagoya, Japan, 1993.

S. Amghar, H. Paugam-Moisy, and J.-P. Royet. Learning methods for odor recognition modeling. In B. Bouchon-Meunier, L. Valverde, and R.R. Yager, editors, Advanced Methods in Artiﬁcial Intelligence, Proc. IPMU’92, LNCS, 682: 361–367. Springer Verlag, 1992.

H. Paugam-Moisy. On a parallel algorithm for back-propagation by partitioning the training set. In N. Giambasi, J.-C. Rault, and M. Ribes, editors, Proc. Neuro-Nîmes’92, 53–66. EC2, 1992.

H. Paugam-Moisy. Optimal speedup conditions for a parallel back-propagation algorithm. In Parallel Processing, Proc. CONPAR’92-VAPP V, L. Bougé, M. Cosnard, Y. Robert, and D. Trystram, editors, LNCS, 634: 719–724. Springer Verlag, 1992.

H. Paugam-Moisy. On the convergence of a block-gradient algorithm for back-propagation learning. In Proc. Int. Joint Conf. Neural Networks, IJCNN’92-Baltimore, III :919–924, 1992.

M. Cosnard, P. Koiran, and H. Paugam-Moisy. Complexity issues in neural network computations. In I. Simon, editor, Latin American Symposium on Theoretical Informatics, Proc. LATIN’92, LNCS, 583: 530–544. Springer Verlag, 1992.

M. Cosnard, J.-C. Mignot, and H. Paugam-Moisy. Implementations of multilayer neural networks on parallel architectures. In 2nd Int. Specialist Seminar on the Design and Application of Parallel Digital Processors, Conf. Publi. 334: 43–47. IEE, 1991.

H. Paugam-Moisy. Selecting and parallelizing neural networks for improving performances. In T. Kohonen, K. M kisara, O. Simula, and J. Kangas, editors, Artiﬁcial Neural Networks, Proc. ICANN-91, I :659–664. North-Holland, 1991.

H. Paugam-Moisy. A parallel tool for evaluating learning capabilities of neural networks. In Hamza, M.H., editor, Proc. IASTED Int’al Symposium in : Machine Learning and Neural Networks, 59–62. ACTA Press, 1990.

Contact

Postal address:

Laboratoire LAMIA -- Dept DMI
Bât. Recherche B, room 323
Université des Antilles
Campus de Fouillole -- BP 250
F-97157 Pointe-à-Pitre
Guadeloupe -- France

Telephone and e-mail:

Helene.Paugam-Moisy-[at]-univ-antilles.fr
hpaugam-[at]-lri.fr

tel. : +590 590 48-3421
fax : +590 590 48-3086

Last update: May 30th, 2017