If you are a Breton researcher and would like to post your expression of interest here, contact us.

Epithelia, Homeostasis, Cellular junctions, Tricellular junctions, Notch signaling, Asymmetric cell division, Tissue plasticity

University of Rennes 1 – Institute of Genetics and Development of Rennes

During development and throughout adult life, cell-fate diversity is in part generated via asymmetric cell divisions. Our team is studying how a cell divides to produce two daughter cells with different identities and how deregulation of asymmetric division is responsible for various types of pathologies, including cancers.Our research themes using invertebrate and vertebrate models combine approaches from genetics, cell biology and soft matter physics.



Mapping enzyme-substrate relationships in the ubiquitin system

University of Rennes 1 – Institute of Genetics and Development of Rennes

Ubiquitin, Proteasome, Protein-protein interaction, Systems biology, Proteomics

Our group investigates at the molecular level the function and regulation of the ubiquitin system. This system controls the activity and stability of the cellular proteome through the covalent modification of proteins by ubiquitin moieties, a process known as “ubiquitylation”. While proteomic experiments have shown that thousands of proteins can be the targeted by ubiquitylation, the great majority of ubiquitylation events remain uncharacterized. In particular, the enzyme that catalyse ubiquitylation have only been investigated for a very limited subset of ubiquitylated proteins. To address this question, we have devised a protein-fragment complementation assay, which enables us to investigate enzyme-substrate relationships in the ubiquitin system at the proteome level. We would like to apply this methodology to systematically investigate the substrates of SCF-family ubiquitin ligases. This will enable us to provide a comparative map of biological pathways targeted by these ubiquitin ligases.


Gwenael RABUT – gwenael.rabut@univ-rennes1.fr

Decoding the molecular and cellular mechanisms governing Drosophila muscle stem cells maintenance and activation.

University of Rennes 1 – Institute of Genetics and Development of Rennes

Keywords: Muscle, Satellite Cells, Stem Cells, Drosophila, Regeneration

Muscle stem (satellite) cells (MuSCs) ensure the functional homeostasis of skeletal muscles as well as their regeneration upon injury. The project goal is to understand how MuSCs sense signals provided by their environment and engage a series of dramatic changes in cellular organization and fate, to repair the muscles. We would like take the advantage of newly designed Drosophila transgenic tools and imaging approaches to track MuSCs in vivo and identify genes regulating their behaviors.

Contact: The muscle Development and Repair team

Hadi BOUKHATMI – hadi.boukhatmi@univ-rennes1.fr

RNA controls in vertebrate development and human disease

University of Rennes 1 – Institute of Genetics and Development of Rennes

Team: Gene Expression and Development

Controls of gene expression exerted at the RNA level are instrumental for the development of vertebrate embryos. This includes alternative splicing, mRNA translation and decay. Our group uses a diversity of in vivo (Xenopus, mice), in cellulo (2D and 3D cultures) and in vitro models. We focus on the contributions of defective controls of RNA regulations in two human pathologies:

(i) Cataract: ocular lens clouding, the leading cause of blindness worldwide. We investigate how RNA regulations impact the organisation of the cytoskeleton, to confer the lens fiber cells their specific shape. Ultimately this peculiar organisation is required for lens transparency. We investigate how defective RNA controls lead to cataract.

(ii) Head and neck squamous cell carcinomas. We investigate how the synthesis of splice isoforms of transcription factors is controlled, and how these different isoforms affect cell properties.


Pr Luc PAILLARD – luc.paillard@univ-rennes1.fr

Solving forebrain disorders and diagnostic impasse with human iPSCs-derived organoids

University of Rennes 1 – Institute of Genetics and Development of Rennes

The research team investigates biological processes that governs brain establishment during early steps of development. We focus on neurodevelopmental disorders to establish their causes at molecular and cellular levels. Our goal is to use the brain organoid technology to study physiopathology of congenital brain disorders related to Sonic Hedgehog-deficiency. It aims to propose a new molecular diagnostic tool to the patient suffering from this pathology by creating a reference transcriptomic identity card of iPSC derived forebrain organoids recapitulating the forebrain spectrum disorder.


Valérie DUPE – valerie.dupe@univ-rennes1.fr

Just transition in French rural territories

University Rennes 2 – LiRIS – Laboratoire interdisciplinaire de Recherche en Innovations Sociétales

Many actors are now involved in ecological and energy transitions in rural areas. The research currently being conducted proposes to study the ways in which these transitions are taking place in these territories, the associated games of actors and power, and the new balances or imbalances that are emerging, in terms of production activities, but also in terms of inequalities in the territories. More specifically, research is already underway within the team on two themes: 1/ the development of the biogas sector in France, and more broadly on the diversification of farmers’ production linked to transitions, and 2/ the definition of a more ecological social protection in France and in Southeast Asia and the way in which social protection actors can position themselves in this context. This work is being carried out by economists in particular, but in close collaboration with other disciplinary fields, notably sociology, geography and political science.


Alexandre BERTHE – alexandre.berthe[at]univ-rennes2.fr


Eco-evolutionary consequences of crayfish invasion on freshwater food webs

Agrocampus Ouest – UMR ESE

Keywords: Biological invasions, Community dynamics, Eco-evolutionary feedbacks, Ecosystem stability, Rapid evolution, Trophic interactions.

Biological invasions often entail deep reorganizations in the trophic structure of native communities, with ramifications to the function and stability of the whole ecosystem. Anticipating and managing these ramifications requires gaining a better understanding of how recipient communities respond to biological invasions. In particular, trophic reorganizations may alter selective pressures and drive rapid evolution of trophic phenotypes of both native and invasive species which, through iterative eco-evolutionary feedbacks, may ultimately change the strength and direction of trophic interactions. To decipher these complex mechanisms, our team uses a combination of high-frequency ecosystem monitoring (flow-through microscopy, multispectral aerial surveys, automated image analysis), molecular and quantitative genetics, mark-recapture and modelling approaches on the field (pond networks), in mesocosms and in the laboratory.



Deciphering microbial metabolism in extreme environments

University of Western Brittany – Biology and Ecology of deep-sea Ecosystems (BEEP)*

Keywords: Metagenomics, Prokaryotes, Hydrothermal/geothermal ecosystems, Sulfur disproportionation

Our team investigates microbial metabolisms in extreme habitats such as deep-sea hydrothermal vents and geothermal hot springs. Currently, we have two main projects. One is focused on the understanding of microbial sulfur disproportionation from the molecular to the ecosystem level. For this project, we would be interested to investigate the history of this metabolism with respect to sulfate-reduction. Our second project is focused on the microbial communities of the geothermal and hydrothermal zones of the Kerguelen archipelago and St Paul Island. These areas are located more than 3000 km from the first inhabited areas and are thus very little anthropized. The microbial communities they host have been the subject of only a few studies and remain largely unknown. We would be interested to uncover metabolic pathways to phenotypic traits of the Bacteria and Archaea from the 30 main sites, using metagenomics.


Karine ALAIN – Karine.Alain@univ-brest.fr –

Tel. (+33) (0)2 98 49 88 53 – K. Alain CV : https://www.univ-brest.fr/menu/recherche-innovation/pages-chercheurs/ALAIN-Karine//CV.cid66547

* European Institute for Marine Studies (IUEM) – BEEP Biology and Ecology of deep-sea Ecosystems UMR6197 CNRS-UBO-Ifremer (formerly LM2E Laboratory of Microbiology of the Extreme Environments)

Formation and regeneration of intestinal microvilli

University of Rennes 1 – Institute of Genetics and Development of Rennes

Genetic, inflammatory or pathogenic disorders, as well as aging, can all induce intestinal microvillus atrophy, causing food malabsorption and diarrhoea associated with severe morbidity. However, intestinal microvillus growth and stability during normal development or in pathological contexts has been mostly investigated in vitro. We take advantage of live super-resolution microscopy and endogenously expressed probes to molecularly characterize C. elegans inherited, acquired or acute microvillus atrophy models and then identify and characterize new regeneration mechanisms.


Grégoire MICHAUX

Structural studies of new toxins targeting the ribosome of the pathogen Mycobacterium tuberculosis

University of Rennes 1 – Institute of Genetics and Development of Rennes

Keywords: RIbosome ; cryo-EM ; toxins ; tuberculosis

Mycobacterium tuberculosis (Mtb) is the causative agent of tuberculosis, the leading cause of death due to a single infectious agent, causing more than 1.5 million deaths per year. To survive in the host, protein synthesis (translation) is tightly regulated in Mtb. This is performed by various quality control factors, including ribosome dependent toxins. Although their cellular functions are still largely unknown, many of these systems are strongly induced in response to stress, e.g. drug exposure, hypoxia, eat shock or DNA damage. We want to focus our studies on the structural aspects of new atypical quality control and TA systems impacting the ribosome. Our main objectives are to get structural insights into the binding of these factors to isolated ribosomes and to solve the structure of the complexes in solution by cryo-electron microscopy (cryo-EM).


Reynald GILLET – reynald.gillet@univ-rennes1.fr

Digital transition and digital twins in the industry

University of South Brittany – Lab-STICC

Keywords: Digital Revolution ; Digital Twins ; Industry 4.0 ; Agro-food Industry ; Decision Support Tools ; Data Analytics/Mining ; Smart Manufacturing

The research team aims to address issues that participate in the acceleration of the digital transition of the French industry and aspires to accompany various economic actors in this process.

The University of Southern Brittany is building a rich and varied ecosystem around the DT, bringing together expertise from different fields to address multidimensional research tracks related to the DT in order to bring it to maturity (Generic models and architectures for DTs, Acceptability and cooperation with DT, proof of concepts and collaborations with industrial partners, etc.).

The aim is to make the Lab-STICC and the Brittany region a pole of competitiveness in France and Europe for DT, by having a wide spectrum of expertise to put to good use for this new transverse technology. This does not prevent us from addressing local challenges, as agro-food related issues, that represents a very important part of the region’s industrial fabric which offers great opportunities for collaboration.


Laboratoire des Sciences et Techniques de l’Information de la Communication et de la Connaissance (Lab-STICC)

Mohammed Adel HAMZAOUI – mohammed.hamzaoui@univ-ubs.fr

ENSIBS website: https://www-ensibs.univ-ubs.fr/fr/index.html

SCAP Factory of the Future platform website: https://www-ensibs.univ-ubs.fr/fr/ecole/equipements-ateliers-2/scap-industrie.html

Coastal territories impacted by Brexit through fishing fleets and fish value chains in north-western France and south-western United Kingdom

University Rennes 2 / UMR CNRS 6590 ESO Espaces et sociétés

Keywords: Fishing geography – Fishing fleets – Fish value chains – Coastal places – Brexit

Within an international team (France, UK, Germany) building the Brexfish project “Fishing Places and Sustainable Transitions”, post-doctoral fellows are welcome to contribute to social science debates and literature on fishing and coastal issues. We are exploring, from theoretical and empirical approaches, how fishing people, places and enterprises in north-west Europe have experienced and responded to Brexit as part of longer-term political-economic and environmental transition. Within this collaboration, we are planning to examine 1/ changes in access to UK offshore waters, focusing on the ports of Lorient (Brittany) and Newlyn (Cornwall), which are home to significant corporate offshore fishing fleets (Scapêche in Lorient and Stevenson’s in Newlyn) and 2/ the effects of changed market access on fish value chains, including export-oriented shellfish fisheries in Cardigan Bay, Pembrokeshire (Wales) and Cornwall, fin fish sales through local fish markets in Cornwall and shellfish and fin fish value chains in Pays de Cornouaille (Brittany) and Port-en-Bessin and related ports (Normandy).


Catherine LAIDIN – catherine.laidin@agrocampus-ouest.fr

Yvon LE CARO – yvon.lecaro@univ-rennes2.fr

Avatars in virtual reality

Inria – HYBRID Team

The research activity of Hybrid team belongs to the field of Virtual Reality and 3D interaction with Virtual Environments. Our objective is to invent novel 3D interactive techniques with virtual environments exploiting both the body and brain of the user. We focus on novel user inputs in virtual reality such as coming from full-body tracking or brain-computer interfaces. Applications of our research program are for industry (virtual prototyping), medicine (surgical simulation, rehabilitation and reeducation), design (architectural mock-up), art or videogames and entertainment. Hybrid was created in January 2013.

We are looking to work further on avatars. Avatars, i.e. digital representations of the users, are more and more present in Virtual Reality/Extended Reality applications. Their use is raising strong research challenges regarding novel devices and immersive displays, multi-sensory rendering algorithms, 3D human-computer interaction techniques, and perceptual models and phenomena.


Anatole LECUYER – anatole.lecuyer@inria.fr

Brain-computer interfaces and neuro-feedback for rehabilitation

Inria – HYBRID Team

The research activity of Hybrid team belongs to the field of Virtual Reality and 3D interaction with Virtual Environments. Our objective is to invent novel 3D interactive techniques with virtual environments exploiting both the body and brain of the user. We focus on novel user inputs in virtual reality such as coming from full-body tracking or brain-computer interfaces. Applications of our research program are for industry (virtual prototyping), medicine (surgical simulation, rehabilitation and reeducation), design (architectural mock-up), art or videogames and entertainment. Hybrid was created in January 2013.

We are looking to work further on brain-computer interfaces and neuro-feedback for rehabilitation. Brain-computer interfaces rely on measuring directly the cerebral activity of the users for interacting with automated systems. Such “mental” interfaces can be used for brain rehabilitation, eg after stroke, by providing direct “neurofeedback”. This is raising strong challenges in terms of real-time brain signal processing (EEG data), and novel human-computer interaction schemes exploiting for instance multi-sensory feedbacks.


Anatole LECUYER – anatole.lecuyer@inria.fr

Impact of sexual reproduction on dispersal and invasiveness of the aquatic invasive plant, Ludwigia grandiflora subsp. hexapetala.

Agrocampus Ouest – UMR ESE

Keywords: Biological invasions, genetic of populations, polyploidy, water primrose, reproductive system, dispersal, Loire basin

Invasive species are good opportunities to understand how para- and peri-patric populations contribute to the biological evolution of organisms. We developed demogenetic models and methods to understand, track and predict the evolution of partially clonal, polyploid populations colonizing new areas. We currently use these mathematical models to infer the ecological and evolutionary processes ongoing on the major aquatic invasive populations of Ludwigia grandiflora subsp. hexapetala colonizing western Europe watersheds. We recently acquired a huge genotyping dataset on 37 populations over the Loire watershed, including temporal and parentage sampling on fewer populations.

We are planning to explore the ecological and evolutionary processes ongoing in these invasive populations, that we can infer from their current genetic diversity and structure. Especially, we are aiming at inferring the reproductive modes in populations and at understanding the spatial structure and dynamics of the colonizing front(s).


Dominique BARLOY – dominique.barloy@agrocampus-ouest.fr

Solenn STOECKEL – solenn.stoeckel@inrae.fr

Brexit, trade and regional differences in France and the UK

Agrocampus Ouest – UMR SMART-LERECO

Keywords: international economics, Brexit, trade agreements, regional disparities, analysis by products, France, UK, gravity model

French regions – and particularly Brittany – are marked by a strong economic role of the agricultural and food sector. Brittany has strong trade relationships with the United Kingdom and is, therefore, particularly vulnerable to Brexit. Any shock that disrupts agricultural and food trade weakens its trade balance.

Our research unit aims to deepen our understanding of the effects of Brexit by exploring regional disparities within France and the UK. The heterogeneity of French regions with respect to their specialization across products, distance to the UK and to alternative markets permits to evaluate the contribution of these factors to the resilience of agrifood industries to external shocks, such as Brexit.

Accounting for the regional decomposition of the British economy enables us to estimate how the non-respect of the new EU-UK trade agreement (of the Northern Ireland protocol) would shape global trade flows, and the dependency on agricultural and food trade.


Marilyne HUCHET – marilyne.huchet@agrocampus-ouest.fr

Angela CHEPTEA – angela.cheptea@inrae.fr

Exploring food fungi for their biotechnological potential through biosynthetic gene cluster expression

University of Western Brittany – Microbial Ecology and Biodiversity Laboratory (LUBEM)

Keywords: Fungi, BGC, secondary metabolites, bioactive compounds

Our team investigates fungal ecosystems in particular in the agri-food but also environnemental contexts.  Filamentous fungi are particularly known to be able to produce natural products with biological activity. We studied various food fungi and have obtained the corresponding genomes. In silico analysis of genome sequences and chromatographic approaches indicate that numerous biosynthetic pathways (BGC) and secondary metabolites (SM) are still unknown. Moreover, some BGC can be silent according to the considered culture conditions. Over the years, we have deciphered several BGC using RNAi but would now be interested in implementing new approaches in our lab for BGC expression, in particular CRISPRa or BGC heterologous expression. The goal is not only to link a given BGC to a secondary metabolite but also to putatively uncover novel metabolites with biotechnological potential, including antimicrobial activity  as an alternative to pesticides (biocontrol) and chemical preservatives, (biopreservation) but also other applications.


Pr. Emmanuel COTON – emmanuel.coton@univ-brest.fr – Tel. (+33) (0)2 90 91 51 41

Multimodal representation deep learning, multimodal adversarial attacks

Inria – IRISA laboratory – LINKMEDIA team

Deep learning based multimodal representations received a lot of attention recently. This has been motivated by their extraordinary ability to facilitate getting insight from the multimedia collections they represent as they convey comprehensive and rich information about elements existing in the real world and captured through various kind of media such as text, images, video, audio, … Deep learning methods, however, are very sensitive to adversarial attacks, to poisoning, to backdoor attacks, etc. Whereas that sensitivity has been rather well explored for medias taken in isolation (adversarial images ; adversarial audio, …), few studies are concerned with multimodal adversarial attacks. Are multimodal representations inherently more robust or are they weaker than uni-modal ones? Should all modalities be attacked to produce an adversarial sample? If not, is there a weakest modality?


Laurent AMSALEG – http://people.rennes.inria.fr/Laurent.Amsaleg/ – laurent.amsaleg@irisa.fr

Agulhas Current Coastal Ecosystem : how does the oceanic circulation variability may impact the first trophic levels of the food chain?

University de Bretagne Occidentale – Laboratory of Physical Oceanography and Remote sensing (LOPS)

The South-African East Coast coastal ecosystem lies landward of the Agulhas Current, one of the most intense Western Boundary Currents in the world’s oceans. The latter follows a southward along-shore path, with speeds commonly exceeding 1.5 m s-1. Between 30° S and 34°S, this path is occasionally disturbed by mesoscale coherent features such as Natal pulses (large solitary cyclonic meanders travelling all the way down to the Agulhas Bank), and semi-permanent cyclonic cold core Durban eddies. The latter will be surveyed during the RESILIENCE cruise in April/May 2022 on board RV Marion Dufresnes. Physical and biological data will be collected at fine scale to characterize the physical processes at play in linking together the very contrasted nutrient-rich coastal waters and the oligotrophic opens-seas waters. Candidates wishing to participate in the analysis of this cruise data with an innovative approach may contact us.


Steven HERBETTE – Team Océan Côtier – steven.herbette@univ-brest.fr

Molecular mechanisms involved in Alterocin production, an antibiofilm protein of Pseudoalteromonas

Université de Bretage-Sud – Laboratoire de Biotechnologies et Chimie Marines (LBCM – IUEM)

Keywords: Mutagenesis, Bacteria, Biofilm, Overexpression

The research thematic of LBCM* is the dynamic of biofilm formation and bioprotection. One axis is focusing on the link between biofilm and virulence. Biofilms are defined as organized bacterial community, embedded in a matrix and capable of persisting in most of environments. It can thus be a reservoir of pathogenic bacteria and a source of antibiotic resistance genes spread. We recently discovered a Pseudoalteromonas strain producing a small antibiofilm protein, we called it Alterocin. We want to understand molecular mechanisms involved in Alterocin production. Alterocin could be a great advance in infectious medicine to minimize biofilm formation of pathogenic bacteria without the use of antibiotics.


Alexis BAZIRE – alexis.bazire@univ-ubs.fr

* Laboratoire de Biotechnologies et Chimie Marines, European Institute for Marine Studies (IUEM) –EA3884. Université Bretagne Sud

Innate immunity and persistence of emerging viruses in the male genital tract

Université de Rennes 1 – IRSET

The transmission of viral diseases is more than ever a major global health concern. The persistence of a number of emerging viruses in the male genital tract and their transmission through semen represents a major challenge for the control of viral dissemination and reproductive health. The UrGenT team (40 people) is a renowned leader in the fields of viral infections and genomics of the human genital tract. One of our main aim is to decipher the mechanisms underlying the persistence of emerging viruses in the testis, with a focus on the role of innate immunity, and to develop predictive models. The team has a unique expertise in ex vivo models of human genital tissues combined with innovative genomic approaches such as single cell RNAseq. The UrGenT team provides a dynamic multidisciplinary research environment for impactful research and career development by talented individuals.


Nathalie DEJUCQ-RAINSFORD – Nathalie.dejucq-rainsford@inserm.fr


Study and modelling of the evaporation kinetics of dairy colloid mixtures during drying

Agrocampus Ouest – STLO lab

Keywords : dairy colloids, drying, evaporation kinetics, modelling of the drying process

During spray drying of a concentrated milk solution, direct observation of particle formation being impossible on an industrial scale (unsuitability of space-time scales), some studies have focused on different types of drying (from the single drop to the thin cell) in a controlled environment and have led to new research perspectives.

Within the UMR STLO, previous work has made it possible to characterise the stages of the drying process in milk protein mixtures and to highlight the competitive influence of the two colloidal components, whey proteins and casein micelles, on the development of these stages and on the final shape of the dry particles. The promising results obtained lead us to address the complex issue of the dynamic exploration of the drying of dairy protein suspensions. Candidates wishing to participate in the investigation of this issue with an innovative approach may contact us.


Cécile LE FLOCH-FOUERE – cecile.lefloch@agrocampus-ouest.fr

Ludovic PAUCHARD – ludovic.pauchard@u-psud.fr (Collaboration with FAST (Fluides, Automatique et Systèmes Thermiques d’Orsay) of Paris-Saclay University)

Composite materials for sustainable maritime transport: racing yachts as opportunities for innovation

University Bretagne-Sud – Institute of Research Dupuy de Lome (IRDL)

Advanced composite materials, such as carbon fibre-reinforced polymers, are becoming popular in the maritime industry, where most of ships are still built with steel.

Avenues are clearly open for lightening ships with these materials, which would allow drastic reduction in fuel consumption and polluting emissions.

Still, several obstacles remain to be overcome to enable them to fully meet the ecological challenges of the maritime sector. Besides, their manufacturing is an arduous task.

Automated manufacturing of composite parts is a promising way for addressing these issues, reducing arduousness of work and generation of manufacturing waste. Moreover, the racing yachts are bold industries that create innovation, later used in maritime transport of passengers and goods.

Our aim is to develop tailored composite materials in terms of specific mechanical properties, additive manufacturing, waste reduction, innovative design and computer simulation.


Pr. Vincent KERVYRIN Vincent.keryvin@univ-ubs.fr


Combining statistical and biochemical approaches for a dynamic modelling of enzymatic digestion

Agrocampus Ouest – IRMAR and STLO labs

Keywords: food digestion; proteolysis; proteomics; dynamics modelling; large data set

A perfect understanding of the digestion process is a prerequisite in order to think food design to specific nutritional targets. However, food digestion is a very complex process, not yet fully understood. The complexity of food digestion lies in the succession of very different phases, and in the complexity of each phase itself. With respect to protein digestion, several proteases are involved, the “operating rules” of which are still incompletely known. Especially, the dynamic dimension of proteolysis is rarely considered, while dynamics is a major characteristic of the digestion process. In an interdisciplinary approach combining the expertise of two research units, in food digestion, protein biochemistry and proteomics on the one hand, and in statistics and data science on the other, we develop innovative methods to investigate this issue. Questions we propose to explore are about inference on peptide filiation throughout digestion based on Mass Spectrometry data and on the influential physicochemical and structural features.


David Causeur – UMR IRMAR CNRS – Institut Agro Rennes-Angers – david.causeur@agrocampus-ouest.fr

Françoise Nau – UMR STLO INRAE-Institut Agro Rennes-Angers – francoise.nau@agrocampus-ouest.fr

Hardware-based protections for a comprehensive approach to counter software and physical attacks on embedded systems

Université Bretagne Sud – Lab-STICC laboratory

Keywords : cybersecurity, hardware design, embedded systems

In the era of Internet of Things (IoT), embedded systems are massively spreading in critical infrastructures such as transport, factories or smart cities. Such devices rely on both small software and hardware components to process and communicate information. They manipulate sensitive data leading to strict security needs. Moreover, these systems are more and more connected to an IT (information technology) network increasing the overall attack surface. This enables new opportunities to breach the global security of infrastructures. Our research group propose, design and evaluate original hardware or mixed (hardware/software) solutions for protecting embedded systems against both software and physical attacks. Software attacks exploit vulnerable code pieces in order to retrieve sensitive information or to perform malicious actions. Contrary to software attacks, physical attacks require a physical access to the target, which is very often easy for embedded systems, in order to retrieve sensitive information through side-channel attacks (SCAs) or to modify its internal state through fault injection attacks (FIAs).


Guy Gogniat – guy.gogniat@univ-ubs.fr

Vianney Lapôtre – vianney.lapotre@univ-ubs.fr

Are Deep Learning Training Accelerators Secure?

Inria – TARAN team

Keywords:  deep learning accelerators, hardware security, fault attacks, countermeasures, fault tolerance

Recent studies show how to use side-channels to guess key parameters or inputs from DL models running on accelerators. For example, input images of a model were estimated from power traces. However, we would like to go further than current practice by studying if private information can be retrieved during the training phase or if it is possible to disrupt the training quality by attacking the accelerators. Our objective is to study training-time hardware attacks and required countermeasures, focusing on fault injections in edge devices and remote side-channel attacks to cloud accelerators.


Olivier Sentieys – olivier.sentieys@inria.fr – http://people.rennes.inria.fr/Olivier.Sentieys/

Spear Phishing Detection through Natural Language Processing and Cognitive Science Methods

IMT Atlantique – LabSTICC – team DECIDE

Spear phishing is a very frequent and effective type of social engineering attack where the victim receives one of more seemingly friendly emails in which the attacker impersonates a colleague or hierarchically higher standing person and asks for a favor or service. We are planning to work on the detection of spear phishing attacks through natural language processing/text mining methods, based on a cognitive model of a generic attack: the persuasion tactics of the attacker, the parameters of perception and the attention failures by the victim. In particular, the context of reception of the attacking messages will be part of the model together with linguistic and paralinguistic information. We are working in an interdisciplinary way (natural language processing, cognitive science, cybersecurity) and with Vade Inc., a company specialized in cybersecurity with an extremely large corpus of email flows.


Yannis Haralambous – yannis.haralambous@imt-atlantique.fr

Philippe Lenca – philippe.lenca@imt-atlantique.fr

How-Provenance Explanations for Efficient Data Mining on Dynamic Datasets

Inria – LACODAM team

Keywords: Rule Mining; Knowledge Bases; How-Provenance

How-provenance annotations are expressions that explain how a particular unit of data, e.g., a query answer, was produced from other data units. In databases, they allow for applications such as dynamic query updates. We are interested in this case because annotations can speed-up the re-evaluation of the scores for patterns mined on dynamic data, incurring also less energy consumption. While we are open to explore this research avenue in a broad sense, i.e., for different types of patterns, we are particularly interested in Horn rule mining on large and dynamic knowledge graphs.


Luis Galárraga – luis.galarraga@inria.fr – http://luisgalarraga.de

Multi-Armed Bandit and Online Learning to Rank

Université de Rennes 1 – IRISA – LACODAM team

Multi-Armed Bandits (MAB) model a large number of applications where an artificial agent chooses an action at each iteration and receives a random reward which depends on this action. The agent’s objective is to maximize the rewards received over time. This setting covers, by example, the choice of news displayed on a web page, or of recommended videos, or even of a journey going from position A to position B.

We are interested by practical MAB algorithms dedicated to specific applications while being theoretically grounded. We especially enjoy settings which boil down to identifying the K most attractive items.


Elisa Fromont – elisa.fromont@irisa.fr – http://people.irisa.fr/Elisa.Fromont/

Romaric Gaudel – romaric.gaudel@ensai.fr – https://sites.google.com/site/romaricgaudel/

How to explain bad decisions taken by neural networks?

University of Rennes 1 – IRISA – LACODAM team

Keywords: Explainable AI, Pattern mining, Neural Network debugging, Conterfactual explanations, Post-hoc explanations

We would like to explain (and potentially act on) the bad decisions (e.g. missclassifications) taken by neural networks. One axis could be to use/learn/construct global (counter-factual) explanations to identify particular problems in the data related to groups of examples (i.e. missclassified examples from the same class) that could, in turn, be used to improve the accuracy of the system.  Another axis could be to use data mining tools to extract meaningful pattern in trained networks (i.e. activation maps per layer) and try to understand if parts of the network can be responsible for particular mistakes.

Elisa Fromont – elisa.fromont@irisa.fr – http://people.irisa.fr/Elisa.Fromont/

Medical applications of speckle field analysis

University of Western Brittany – OPTIMAG laboratory

Speckle field analysis is an effective tool for the study of microscopic processes in biological media. Speckle is generated when coherent light is scattered by a rough surface at the wavelength scale (or much larger), or by suspended particles in a liquid medium. The resulting speckle images show a random spatial distribution of intensity with bright and dark spots, resulting from interferences of the scattered light. The dynamic and spatial characteristics of the observed phenomenon are directly related to the physical properties of the studied sample. Our research team (OPTIMAG – Optics team) has already exploited the potential of this method in various fields, such as agriculture, for the monitoring of fruit ripening or the optical characterisation of a biopesticide, odontology, for the early diagnosis of dental erosion or for monitoring the hardening kinetics of dental cements, or oenology, for determining the origin of a wine. Our purpose is now to develop medical applications of speckle field analysis in collaboration with clinicians from our Institute (IBSAM – Brest Institute in Health, Agronomy and Materials), for example for the detection of diseases causing changes in the constitution of biological fluids (human blood coagulation, semen hyperviscosity, …).


Guy LE BRUN – Guy.LeBrun@univ-brest.fr

Machine learning-based biofouling development prognosis in tidal stream turbines

University of Western Brittany – Institut de Recherche Dupuy de Lôme

In a context of marine renewable energies harvesting, biofouling can easily cause obstructions in marine renewable energy converters (i.e. tidal stream turbines) and/or increase the weight and drag, thus significantly affecting the device performance. Two potential performance issues for tidal stream turbines are blades roughening due to impact, cavitation or scour due to particulates, and blades fouling by marine growth. Biofouling is therefore a major engineering concern, influencing offshore structures loading by increasing structural elements size, increasing drag and inertia coefficients, as well as increasing the structural weight. Furthermore, biofouling may damage protective coating and interfere with sensitive areas necessary for monitoring and maintenance. Care will therefore be required in this aspect of turbine maintenance, depending on factors such as areas of operation, sea temperature, as well as other physico-chemicals factors.

Our research team started to work on the detection of tidal stream turbine biofouling using machine learning-based approaches for maintenance scheduling purposes. The biofouling issue is expected to be addressed using electrical terminals of the tidal turbine generator.


Mohamed BENBOUZID – Mohamed.Benbouzid@univ-brest.fr

Hydrogen storage of carbone based or derived monolayers supported on metallic substrate.

University of Western Brittany – CEMCA Lab

Keywords: Electronic structure simulation, DFT, Neuronal Networks, statistical physics analysis

Monolayers based on carbon, boron or other light elements are good candidate to store hydrogen molecules that could latter served in fuel cells. Most of the systems must be doped in order to reach a reasonable gravimetric content of hydrogen molecules that meet the standard defined by the American department of energy. The weight of hydrogen molecule with respect to the overall system must exceed 5.5 at weight percentage and the adsorption of the molecule must enter a narrow window of energy in the range 0.4-0.15 eV. Lots of publications have recently issued on this research item and one is amazed by the abundance of promising systems with potentially high hydrogen molecule content. Though promising these systems are hardly explored with respect to the addition of pressure or of external constraints. With the help of large computing facilities and the possibility of a “secondment” at the University of Valladolid we would like to recruit a fellow candidate highly talented and experienced with electronic structure simulations and the use of neuronal networks.


Alexandre LEBON – alexandre.lebon@univ-brest.fr

Study of marine trace metal cycles

University of Western Brittany – LEMAR lab

At the Laboratoire des Sciences de l’Environnement MARin (Plouzané), we are interested in the study of marine trace metal cycles. Some of these metals being key nutrients for the phytoplankton, they strongly influence the structure of marine ecosystem and the carbon drawdown through the biological pump. In this context, we perform many laboratory experiments to study processes involved, using or state-of-the art cleanrooms. We also participate to many ocenaographic cruises in various parts of the worlds oceans (Southern Ocean, North Atlantic), many of which falling under the umbrella of the international GEOTRACES program.



Hélène Planquette – helene.planquette@univ-brest.fr

Advanced study of anisotropy and instability of exotic or active architected materials

University of South Brittany – Institut de Recherche Dupuy de Lôme

Keywords: Structural mechanics, strength of materials, instabilities, multi-scale behavior, symmetries

I am interested in developing efficient tools to study complex materials with internal microstructure such as composites, cellular, architected or hierarchical materials, using new theoretical methodologies and advanced numerical simulations.


Christelle COMBESCURE – christelle.combescure@univ-ubs.fr