International initiatives

ANR Project Keops

  • contact: Thierry Vieville

Title: Algorithms for modeling the visual system: From natural vision to numerical applications

Overview. The design of artificial sensory systems has evolved in parallel to our knowledge on biological systems. Although the field of engineering can take advantage of sensory biological solutions discovered in nature, only recently some bioinspired solutions for visual applications have emerged. In the last years and with the advent of new neural acquisition methods and theoretical frameworks a better understanding of the neural coding process - from physical signals to neural networks - has emerged (Simoncelli and Olshausen 2001, Hemmen and Sejnowski 2006). More particularly, a recent description in the retina of non-standard ganglion cells types, beside a complex repertoire of standard ganglion cells responses in front of natural stimulus convey important questions about the real, early processing capacity of the retina. This leads to revisit both the neural coding of the information the eye is sending to the brain, and also sheds light to engineering applications from the understanding of such encoding, as detailed in the sequel. At the modeling level, retinal cells are mainly formalized using a LN (Linear spatio-temporal filtering followed by a static Non-linear transduction), while an important fraction of non-standard cells response cannot be represented in such a model class. This is thus a challenge to develop an innovative formalism that takes such complex behaviors into account, with such immediate applications as new dynamical early-visual modules. Proposing new innovative bioinspired formalisms in order to perform dynamical visuo-perceptual tasks adapted to natural environment is a main goal of this project, with a special focus to scenes including complex visual motion interacting with light.
The project is a cooperation between the University of Nice (France), the University of Valparaiso (Chile), the Pontifical Catholic University of Chile in Santiago de Chile and CORTEX.

INRIA associate team Cortina

  • contact: Frédéric Alexandre

Title: CORtex and reTINA modeling from an engineering and computational perspective

Overview. Much progress has been made in the last decades in understanding the basic organization and function of the nervous system in general. Contributions to this end have come from various domains including computational neuroscience and numerical science of the information in general. The goal of this associate team is to combine our complementary expertise, from experimental biology and mathematical models (U de Valparaiso and U Federico Santa-Maria) to computational neuroscience (CORTEX and NEUROMATHCOMP), in order to develop common tools for the analysis and formalization of neural coding and related sensory-motor loops. Recording and modeling spike trains from the retina neural network, an accessible part of the brain, is a difficult task that our partnership can address, what constitute an excellent and unique opportunity to work together sharing our experience and to focus in developing computational tools for methodological innovations. To understand How the neural spike coding from natural image sequences works we are addressing the following issues: How visual signals are coded at earlier steps in the case of natural vision? What are their functions? What are the computational “coding” principles explaining (in artificial or biological system) the statistical properties of natural images? We wish to advance our actual knowledge in natural and artificial visual signals processing and apply it to the field of education; to foster better capacities for learning and memory; sensory prosthesis design, to will help unpaired sensory persons to sense the world and physical rehabilitation, among others. In the context of the cooperation between the INRIA and Chile our methodology and questions are transversal and can be used and potentiated several field and applications.
The association is a cooperation between Cortex and the University of Valparaiso (Chile).
For more information: http://cortex.loria.fr/pub/research/cortina.html

INRIA associate team CorTexMex

  • contact: Bernard Girau

Overview. Computational vision aims at extracting useful context-dependent information from images. In spite of the progresses in computer vision, some visual tasks cannot be performed satisfactorily due to the over simplified nature of classical models compared to the intrinsic complexity of the environment. To alleviate this problem, one of the main current research lines is directed to using biologically plausible models of visual perception by understanding, modeling and simulating the mechanisms that underly neural processes in the brain. Moreover, implementing such models in artificial systems may help to achieve better, faster and more efficient embedded systems for visual perception. Nevertheless, the high computational cost of these models usually exceeds the time-multiplexed bounded computational resources of conventional systems. A solution lies in alternative hardware/software based processing architectures, supporting biological realism and providing the large scale computational resources to satisfy application constraints. The CorTexMex associate team focuses on the analysis, methods and techniques for the embedded implementation of bio-inspired connectionist processing for visual perception on reconfigurable devices under a hardware/software approach. The main goal is to provide methods able to handle the massive distribution and the connection complexity of these models, as well as their specific recurrent differential computations. Another goal is to provide bio-inspired connectionist processing models for visual perception to be embedded and directly integrated in perception-action loops. In this context we mainly address two biologically inspired models for visual tasks: spiking networks and dynamic neural fields. In the long term, this joint research is expected to facilitate the cooperation between computational neuroscientists and computer scientists to develop fully functional biologically inspired vision systems in autonomous robotics.

STIC-AmSud project BAVI

  • contact: Bernard Girau

Overview.This project lies in the field of audio-visual information integration. The approach is based on the derivation of distributed models from neurophysiologic studies of motion perception in the human brain, and takes advantage of advanced methods for audio-visual information integration and visual animation. We aim at improving audio-visual information integration by using bio-inspired neurocomputational models that are characterized by their massive distribution and their robustness, to extract visual patterns of phoneme-related face motions. In addition, we aim at improving speech-driven face animation by relating acoustic signals, face motion features and visual animation techniques. Both goals join together towards the definition of bio-inspired models for audio-visual integration that derive from an implicit cortical sensory (audio/visual)-motor (animation) loop. To reach these goals, we combine the skills of three teams that are respectively specialized in cortex-inspired models of visual perception, audio-visual information integration and visual animation.

STIC-AmSud project BCI

  • contact: Laurent Bougrain

Overview. This proposal aims to develop computational intelligence techniques for pattern recognition of graphic elements (e.g. event-related potential, auditory evoked potential, k-complex, spindle) included in electro-encephalographic signals. More precisely, we want to develop adaptive computational intelligence techniques based on artificial neural networks, support vector machines and classical data analysis techniques to robustly detect evoked potentials in a single trial from noisy and multi-sources electro-encephalographic signals. The results of this work will be a comparison of the most powerful techniques developed and a complete procedure specifying the acquisition system parameters, the preprocessing techniques and a robust learning technique able to faster detect evoked potentials for brain-computer interfaces. This methodology will be available for other problems such as, in the medical domain, auditory evoked potential detection for automated newborn hearing screening and k-complex detection for automatic sleep scoring.


European initiatives

FP7-ICT project NEUROCHEM

  • contact: Dominique Martinez

The European project NEUROCHEM explores biologically inspired computation for chemical sensing, incollaboration with the University of Barcelona, the royal institute of technology (Sweden), INRA (Paris),the university of Manchester, the University Pompeu Fabra (Spain), CNR-IMM (Italy) and the University of Leicester.

Overview. Biological olfaction outperforms chemical instrumentation in specificity, response time, detection limit, coding capacity, time stability, robustness, size, power consumption, and portability. This outstanding performance is due to the unique architecture of the olfactory pathway, which combines a high degree of redundancy, an efficient combinatorial coding along with unmatched chemical information processing mechanisms. NEUROCHEM will develop novel computing paradigms and biomimetic artefacts for chemical sensing taking inspiration from the biological olfactory pathway. This project proposes to build computational models of its main building blocks: olfactory receptor layer, olfactory bulb, and olfactory cortex. Complexity of the biological inspired models will go through an abstraction stage in which their processing capabilities are captured by algorithmic solutions. The project will demonstrate this approach building a biomimetic demonstrator featuring a large scale sensor array mimicking the olfactory receptor neuron layer, where abstracted biomimetic algorithms will be implemented in an embedded system that will interface the chemical sensors.


National initiatives

ANR project PHEROSYS

  • contact: Dominique Martinez

Overview. This collaborative project in systems Biology (project ANR-BBSRC SysBio) with INRA (Paris, FR) and the Universityof Sussex (UK) explores olfactory coding in the insect pheromone pathway through models and experiments.More information available at http://www.informatics.sussex.ac.uk/research/projects/PheroSys/index.php.

ANR project MAPS

  • contact: Frédéric Alexandre

Overview. This collaborative project with INCM in Marseille, UMR Perception and Movement in Marseille and LIRISin Lyon aims at re-examining the relationship between structure and function in the brain, taking into accountthe topological (spatial aspects) and hodological (connectivity) constraints of the neuronal substrate. We thinkthat those constraints are fundamental for the understanding of integrative processes, from the perception levelto the motor level and the initiation of coordinated actions.

ARC Amybia

  • contact: Bernard Girau

Overview. Our regular collaborations with researchers from the Maia team has shown that we share common computation paradigms based on massively distributed and local models that are inspired by biological systems. This has led us to join our efforts in an original collaboration within the Amybia project led by Nazim Fatès (Team MAIA), together with Hugues Berry who works on similar models by exploring a bio-inspired approach to propose challenging paradigms for spatial computing within the Alchemy team. This collaboration is also linked with our hardware implementation activities, since it has resulted in an embedded implementation of a biological inspired model for the decentralized gathering of computing agents, as well as in a blocksynchronous implementation of the environment of this model to study its phase transition properties.

ARC Maccac

  • contact: Thierry Viéville

Overview. Since neuronal information processing is related to the brain bio-electrical activity, measured by current neuroimaging techniques at different time and space scales, from neurons to the brain as a whole (e.g.LFP, ECoG, EEG, MEG), the analysis of such complex data coming from these measurements requires the parallel development of suitable models. Namely, these models have to be, on the one hand, close enoughto phenomenology, taking into account the various type of bio-electrical activity and their scales relations, inorder to propose a coherent representation of information processing in the brain (from neurons to neuronalpopulations, cortical columns, brain area, etc). On the other hand, these models must be well posed andanalytically tractable. This requires a constant interaction between neurobiology, modeling and mathematics.In this spirit, this project, directed by Bruno Cessac (INRIA Team NEUROMATHCOMP), aims to tackle the followingquestions: (i) Mesoscopic modeling of cortical columns, bifurcations, and imaging. (ii) Statistical analysis ofspike trains. The CORTEX team brings its computer science expertise, mainly regarding the question (ii) andthe OI modality regarding the question (i).

DGE Ministry grant COMAC “Optimized multitechnique control of aeronautic composite structures”

  • contact: Laurent Bougrain

Overview. The goal of this three-years project is to develop a powerful system of control on site, in production and in exploitation, of aeronautical pieces made of composite. It takes up the challenge of the precise, fast and local inspection on composite pieces of aeronautical structures new or in service by using techniques of nondestructive control more effective and faster to increase the lifespans of the structures of planes. This project requires a decision-making system including fast methods of diagnostic based on several optical technics as non-destructive control.)

Action Modeling, Simulation and Interaction of the CPER

  • contact: Hervé Frezza-Buet

Overview. In the framework of the Contrat de Projet État Région, we are contributing to the axis MIS (Modeling,Interaction and Simulation) through the project InterCell whose goal is to study massive cellular computations in an interactive framework National initiatives

Project of the CNRS NeuroInformatics program on olfaction

  • contact: Dominique Martinez

Overview. The project "Olfactory coding" (2008-2009) from the CNRS program "Neuroinformatics" with the CNRS UMR 5020 (Lyon) explores the role of spike timing in olfactory coding.

Project of the CNRS NeuroInformatics program on reinforcement learning

  • contact: Frédéric Alexandre

Overview. In this collaboration with the MAIA team, Supelec Campus de Metz and the Interative and CognitiveNeuroscience Centre in Bordeaux, we are developing bio-inspired reinforcement learning procedures, on thebasis of experimental data from behavioral recordings in rats.

Project of the CNRS NeuroInformatics program on neural coding in the retina

  • contact: Axel Hutt

Overview. How the interplay between sensory receptors and their primary interneurons becomes a perception is largely unclear, yet it represents an important step in sensory coding and perception. The codification of sensory information-the steps from sensory transduction to processing in the brain- displays great similarity in its principles and underlying elements between the senses. The project aims at understanding the role of distributed neuronal network activity in sensory perception in vertebrates including human beings, using physiological to formal neural network approaches. An integrative approach of researchers from different areas to the common problems of sensory transduction, transmission, and codification should yield significant synergy, and bring us closer to the ultimate goal of a full and detailed comprehension of our senses. The project is a cooperation between the University of Nice (France), the University of Valparaiso (Chile), the Pontifical Catholic University of Chile in Santiago de Chile and CORTEX.

Project of the CNRS NeuroInformatics program on cortical signals to control a two-finger robotic hand driven by artificial muscles

  • contact: Laurent Bougrain

Overview. Nowadays, the understanding the control of manual dexterity in primates can be reached. Over the last twenty years, thanks to improved techniques for intra-cranial recordings, several advances have been obtained in particular to predicting the direction of movement of the upper limb. Recent work has shown that it is possible to predict from brain data the flexion and the strength of fingers. The main objective of this project is to study the control of two anthropomorphic fingers (index finger and thumb) through intra-cortical signals recorded in the monkey during grasping movements (precision grip), forecasting both the finger position and the electromyographic activity (EMG) of the muscles involved in the movements of these two fingers. The project aims to (i) acquire high-quality recordings using an array of 96 micro-electrodes, (ii) improve our experimental site for the grasping, and (iii) evaluate new modelings. This project is a cooperation between the University of Paris V, the Mediterranean Institute for Cognitive Neuroscience (INCM) and the research team CORTEX.

Project of the CNRS NeuroInformatics program : Olfactory rhythms

  • contact: Thomas Voegtlin

Overview. In the rat olfactory bulb, two distinct oscillations of the local field potential (LFP) are observed during the respiratory cycle. A gamma-range oscillation (60Hz) is observed during the transition between inhalation and exhalation, followed by a beta-range oscillation during exhalation. This alternation between beta and gamma is prominent in the anesthetized animal, but weaker in the wake state. Computational studies have suggested that gamma oscillations are generated by the interplay between excitatory mitral/tufted cells, and inhibitory granule cells. However, it is not known what mechanism is responsible for the highly stereotyped switch between beta and gamma frequencies during respiration. The project aims at investigating the origin of this frequency switch. We propose three hypotheses, that are inspired by the current knowledge of the olfactory bulb, and by general principles that govern the behaviour of oscillatory systems. These hypotheses will be tested by experiments and computer models, in order to retrieve the origin the frequency switch.