Offre en lien avec l’Action/le Réseau : – — –/Doctorants

Laboratoire/Entreprise : Sorbonne Universite – Institut des Systèmes intell
Durée : 36 mois
Contact : patrick.gallinari@sorbonne-universite.fr
Date limite de publication : 2022-09-30

Contexte :
Deep Learning is beginning to be explored for scientific computing in domains traditionally dominated by physics models (first principles) like earth science, climate science, biological science, etc. It is particularly promising in problems involving processes that are not completely understood, or computationally too complex to solve by running the physics inspired model. The direct use of pure machine learning approaches has however met limited successes for scientific computing. Hence, researchers from different communities have started to explore (i) how to integrate physics knowledge and data, and (ii) how to push the limits of current ML methods and theory; two challenging directions.

Sujet :
The global objective of the thesis is the development of new models leveraging observation or simulation data for the modeling of complex spatio-temporal dynamics characterizing physical phenomena such as those underlying earth-science and climate observations. The classical modeling tools for such dynamics in physics and applied mathematics rely on partial differential equations (PDE). Despite their successes in different areas, current ML based approaches are notably insufficient for such problems. Using ML for physics raises new challenging problems and requires rethinking fundamental ML ideas.

Three main research directions will be explored:

– Hybrid systems – Integrating Physics and Deep Learning. In many situations, there is available some prior physical knowledge provided by PDEs for characterizing the underlying phenomenon. A key issue is then how to combine this prior knowledge with information extracted from the data.

– Domain generalization for deep learning as dynamical models. Explicit physical models come with guarantees and can be used in any context (also called domain or environment) where the model is valid. This is not the case for DNNs, and we have no guarantee that they can be extrapolated to new physical environments.

– Learning at Multiple Scales. Modeling dynamical physical processes often requires considering multiple spatio-temporal scales. For example in climate, global phenomena are influenced by dynamics operating at a smaller scale. Similar problems occur e.g. in computational fluid dynamics. Learning at different scales is an open issue in ML.

Motivating problems and applications will come from climate science.

Profil du candidat :
Computer science or applied mathematics, Good technical skills in programming.

Formation et compétences requises :
Master in computer science or applied mathematics, Engineering school. Background and experience in machine learning. Good technical skills in programming.

Adresse d’emploi :
Sorbonne Université (S.U.), Pierre et Marie Campus in the center of Paris. The candidate will integrate the Machine Learning and Deep Learning for Information Accesss team at S.U. at the ISIR lab. (Institut des Systèmes Intelligents et de Robotique).

Document attaché : 202203010947_2022-PhD-Description-ML-Physics.pdf

Offre en lien avec l’Action/le Réseau : – — –/Doctorants

Laboratoire/Entreprise : LAMSADE
Durée : 3 ans
Contact : kbelhajj@googlemail.com
Date limite de publication : 2022-04-30

Contexte :
Le monde des services financiers est depuis longtemps l’un des précurseurs de l’innovation digitale à travers l’informatisation de ses processus, de ses outils et de ses services. La période récente voit l’émergence de nombre de nouvelles technologies faisant évoluer les pratiques, et notamment la technologie Blockchain qui a le potentiel de totalement remodeler les services financiers dans le monde de la banque de financement et d’investissment (par exemple activités d’émissions sur les marchés, de financements, de gestion et de couverture des risques).

Sujet :
Dans ce contexte, BNP Paribas CIB en collaboration avec l’Université Paris-Dauphine explore les opportunités et impacts de la tokenization liés à la technologie Blockchain. Une certaine expertise a étét déjà aquise basée sur des développements concrets, en prenant en compte les aspects technologiques, légaux et réglementaires. Il reste cependant un grand champ d’incertitudes liées à cette technologie, et notamment l’interopérabilité entre les différents protocoles. A ce stade, il est clair que la consolidation technologique n’a pas eu lieu, et n’aura probablement pas lieu avant plusieurs années, et dans certains domaines seulement. Par exemple, la tokenization de « cash » (via les acteurs centraux par exemple) et la tokenization de titres financiers seront probablement sur ces bases technologiques différentes. De la même manière, la gestion d’espaces publics (e.g. Ethereum) et d’espaces privés sera instrumentale dans la capacité à gérer la valeur financière dans ce nouveau monde.

Les travaux s’orienteront donc majoritairement sur l’interopérabilité technologique, que ce soit via les divers langages de smart contrats, des techniques de cryptographie, la gestion d’« incentives » d’un écosystème, les tokenomics etc. afin d’assurer l’unicité digitale d’un univers à l’autre de manière sécurisée et irréfutable

Les orientations données à la thèse pourront être choisies au prisme des intérêts de recherche et de leur caractère scientifiquement novateur, des appétences du candidat, via une discussion avec l’encadrant CIFRE et le directeur de thèse.

Profil du candidat :
Vous êtes diplômé(e) d’un bac+5 et souhaitez vous orienter vers un doctorat en Sciences de l’Informatique.
Vous avez également des appétences pour les nouvelles technologies et les leviers digitaux. Vous parlez couramment français et anglais et vous maîtrisez un langage de développement type Java. Des connaissances sur les bases de la cryptographie, les technologies Blockchain (layer 1 et/ou 2) ainsi que les langages de smart contrats les plus communs (e.g. Solidity pour Ethereum/Polygon, SmartPy pour Tezos, DAML, Hyperledger Fabric…), les technologies d’interopérabilités type Cosmos ou Polkadot sont un plus.

Formation et compétences requises :
Votre adaptabilité et votre capacité à collaborer sont des atouts essentiels. Ajoutez à cela votre capacité à communiquer, votre capacité de synthèse ainsi que votre rigueur pour finir de nous convaincre.

Enfin, nous attachons une importance particulière à ce que nos futurs collaborateurs agissent au quotidien avec responsabilité éthique et professionnelle.

Dans un monde qui change, la diversité, l’équité et l’inclusion sont des valeurs clés pour le bien-être et la performance des équipes. Chez BNP Paribas, nous souhaitons accueillir et retenir tous les talents sans distinction : c’est ainsi que nous construirons, ensemble, la finance de demain, innovante, responsable et durable.
À tout moment pendant le processus de recrutement, les informations figurant sur votre CV et vos données d’identification pourront être vérifiées par un prestataire extérieur mandaté par BNP Paribas.

Indications complémentaires en vue du recrutement
Date de recrutement : à partir de septembre 2022
Rémunération: compétitive à l’échelle d’un consultant BNP.
Financement : bourse de thèse CIFRE BNP Paribas CIB
Environnement, une équipe BNP bien affirmée en matière de recherche et développement qui compte une dizaine de doctorants Cifre.
Laboratoire académique : LAMSADE, Université Paris Dauphine – PSL

Adresse d’emploi :
LAMSADE, Université Paris Dauphinen, Paris, et
BNP, Paris

Les personnes intéressées sont invitées à envoyer leur CV, relevés des trois années dernières (et de cette année si applicable), ainsi que la lettre de motivation à Daniela Grigori (daniela.grigori@lamsade.dauphine.fr) et Khalid Belhajjame (kbelhajj@gmail.com)

Document attaché : 202202270954_AnnonceTheseCifre-Dauphine-BNP.pdf

Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : LIFAT EA 6300 (lifat.univ-tours.fr)
Durée : 3 ans
Contact : donatello.conte@univ-tours.fr
Date limite de publication : 2022-02-24

Contexte :
The thesis will be part and funded (gross salary: 2 000 € approximately) under the ANR project CodeGNN (http://www.normastic.fr/projet-anr-codegnn/)
The PhD could start around October 2022

Supervisors and contact
Donatello Conte (University of Tours, France) donatello.conte@univ-tours.fr
Sébastien Bougleux (Université de Caen Normandie, France) bougleux@unicaen.fr
Nicolas Ragot (University of Tours, France) nicolas.ragot@univ-tours.fr

Ph.D. Position (10/2022): Learning Spatio-temporal data by graph representations

Sujet :
In many application domains like action recognition or prediction, video segmentation, traffic forecasting or anomaly detection in brain activity signals, time-varying data are frequently represented by graphs. Two main representations are commonly considered: a temporal sequence of graphs or a spatio-temporal graph connecting graph nodes through time. While there is a solid literature on data analysis based on such representations, the domain has strongly evolved over the last 5 years with the advances in deep learning on Graph Neural Networks.
Such methods have been less investigated for time-varying graphs, particularly when both the graph structure and the data attached to this structure are varying.
We can distinguish two main models: Recurrent Neural Networks (RNN) combined with spatial convolutions rely on the sequential representation [1, 2, 3]; or Graph Convolutional Networks alternating temporal and spatial convolutions [4, 5, 6].
The aim of this thesis is to:
1. Study new representations for spatio-temporal graphs: we want to investigate some new representations in two main directions: representing temporal data as attributes of nodes and edges, and representing temporal data as edge connections between spatial positions represented by nodes at different times.
2. Propose new Neural Network architectures for data represented by this kind of graphs: we want to propose adapted convolutions, decimation and pooling, and study the definition of a recurrent neural network that operates directly in the space of the graphs (for example generating new graphs). One direction of study will also be the Spatial- Temporal Graph Attention Networks (STGAT [7]) and Graph Transformer Networks (GTN [8]).
3. Program these models (in Python), and compare them to the state-of-the-art on standard datasets for different applications, in particular, skeleton-based gesture recognition.

References
[1] A. Jain, A. R. Zamir, S. Savarese, and A. Saxena, “Structural-rnn: Deep learning on spatio- temporal graphs,” CoRR, vol. abs/1511.05298, 2015
[2] C. Si, W. Chen, W. Wang, L. Wang, and T. Tan, “An attention enhanced graph convolutional lstm network for skeleton-based action recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1227–1236, 2019
[3] Li, M., Chen, S., Zhao, Y., Zhang, Y., Wang, Y., & Tian, Q. (2020). Dynamic multiscale graph neural networks for 3d skeleton based human motion prediction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 214-223).
[4] B. Yu, H. Yin, and Z. Zhu, “Spatio-temporal graph convolutional neural network: A deep learning framework for traffic forecasting,” CoRR, vol. abs/1709.04875, 2017.
[5] L. Shi, Y. Zhang, J. Cheng, and H. Lu, “Skeleton-based action recognition with directed graph neural networks,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 7912–7921, 2019.
[6] Chen, T., Zhou, D., Wang, J., Wang, S., Guan, Y., He, X., & Ding, E. (2021). Learning Multi-Granular Spatio-Temporal Graph Network for Skeleton-based Action Recognition. arXiv preprint arXiv:2108.04536.
[7] Kong, X., Xing, W., Wei, X., Bao, P., Zhang, J., & Lu, W. (2020). STGAT: Spatial-temporal graph attention networks for traffic flow forecasting. IEEE Access, 8, 134363-134372.
[8] Yun, S., Jeong, M., Kim, R., Kang, J., & Kim, H. J. (2019). Graph transformer networks. Advances in neural information processing systems, 32.

Profil du candidat :
– Master degree in Computer Science, Applied Mathematics, Data Science, or similar.

Formation et compétences requises :
– strong background in computer science and maths
– experiences in neural networks, deep learning, Python programming,
numerical analysis will be privileged
– knowledge in video and image analysis would be appreciated
– good communication skills and reporting, autonomy and curiosity

Adresse d’emploi :
LIFAT, 64 Avenue Jean Portalis, 37200 Tours

Document attaché : 202202231030_PhD_Thesis_Proposal_SpatioTemporalGraphs.pdf

Offre en lien avec l’Action/le Réseau : – — –/Doctorants

Laboratoire/Entreprise : Institut Fresnel
Durée : 3 ans
Contact : andre@fresnel.fr
Date limite de publication : 2022-02-24

Contexte :
With the rise of Internet-of-Things (IoT) applications and the need for massive connectivity, future 6G networks should meet the demands for the global access to high-speed Internet [1]. One of the envisaged solutions consists in deploying non-terrestrial networks such as networks of satellites or microsatellites in the low Earth orbit (LEO). Such satellites have a much lower manufacturing and launch costs than the traditional satellites, such as those placed in the geostationary orbit. Such very high-throughput satellite (VHTS) networks will be able to meet the future substantial data traffic requirements [1,2]. The specificity of these satellites (or microsatellites) is that they have limited capacities and resources (energy, computing, etc.). However, they are more flexible in terms of resource management, such as power and bandwidth allocation. Another particularity of such networks is the irregular distribution of users (on the Ground) and the variability of connections and, therefore, the data traffic over time. This calls for energy efficient and high-speed connectivity solutions for inter-satellite and satellite-to-ground links. In particular, the use of laser communications or free-space optics (FSO) technology promises high rate and secure data transmission over very large distances [3].

Sujet :
In practice, the establishment of such links is associated with several challenges in terms of (a) link availability/reliability and (b) resource management at the satellite. Indeed, the irregular distribution of users (on the Ground) and the variability of data traffic during the day appeal for the design of efficient architectures with flexible resource allocation according to the requested traffic [6].

(a) The first objective is to propose advanced transmission techniques to establish high-speed communication links with high-reliability between microsatellites or between a microsatellite and a Ground station [7]. These solutions must in particular take into account the atmospheric
channel and the vibrations of the payloads, which can cause significant pointing errors (i.e., misalignment between the transmitter and the receiver) [4]. This first step includes the modeling of optical communication channels and will be carried out in collaboration with the
University of Edinburgh.

(b) In a second step, machine learning-based mechanisms will be designed for performing automated resource allocation in order to increase the capacity of satellite-Earth links [8-9]. This will exploit the flexibility of microsatellites in terms of resource management, such as
power and bandwidth allocation.

For more details, see the attached file.

Profil du candidat :
A solid background in signal processing is an important asset. An experience or training in digital communications is also very welcome. The candidate must have a very good English language proficiency (oral and written expression) and be keen for short-term stays in partner laboratories.

Formation et compétences requises :
Master/engineering school in signal processing, telecommunication, data sciences, statistics, mathematics, computer sciences …

Adresse d’emploi :
52 Av. Escadrille Normandie Niemen, 13013 Marseille

Document attaché : 202202221036_Thesis-SatCom-FSO-English(1).pdf