1) Declarative Constraints for privacy-friendly ML Systems
Link: https://project.inria.fr/hyaiai/files/2020/09/postdoc_consfair.pdf
Contacts: Jan Ramon and Elisa Fromont

2) Understanding and Explaining Complex Systems with Causal Relationships
Link: https://project.inria.fr/hyaiai/files/2020/01/post-doc-causal.pdf
Contacts: Michèle Sebbag, Marc Schoenauer, Alexandre Termier and Elisa Fromont

3) Interpretable causal discovery
Link: https://project.inria.fr/hyaiai/files/2020/09/postdoccausalmag.pdf
Contacts: Jan Ramon and Marc Schoenauer

We are looking for a highly motivated computing scientist with a strong academic and scientific background in Machine Learning and a real interest in medicine and oncology. Skills requested: – A PhD degree in Computer Science, Applied Mathematics or Electrical Engineering with an emphasis on Machine Learning – At least 3 years project experience in AI/ML/DL/MI (included PhD) – Good publication record – Excellent oral and written communication skills Work environment: The candidate will integrate a multidisciplinary team composed of physicists, physicians, engineers and biologists, with dedicated research in radiomics and artificial intelligence for RT +/- IO purposes. U1030 is part of Gustave Roussy Cancer Campus which is a world class cancer center leader in immunotherapy, precision medicine and cancer biology. The brachytherapy department is specialized in the management of gynecological cancers. It treats 200 patients with cervical cancer per year. It is particularly active in the development of combination strategies with targeted molecular therapies and immunotherapy (phases I and II). The department of Pathology has a strong expertise in gynecological pathology and has developed an experience in immuno-oncology.
Conditions of employment: the position is offered for 24 months, starting as soon as possible. Salary: depending on experience

Apply: applicants should send a cover letter summarizing past experiences and research interests, a CV including a list of publications, and contact information for two references to Dr Charlotte Robert, Associate Professor at Paris Saclay University.

Apply: applicants should send a cover letter summarizing past experience and research interests, a CV including a list of publications, and contact information for two references to Pr Eric Deutsch, chair of the radiation oncology department Gustave Roussy cancer campus and INSERM U1030 unit eric.deustch@gustaveroussy.fr and Maria Vakalopoulou, Associate Professor at CentraleSupélec maria.vakalopoulou@centralesupelec.fr

Objectives: Considering, we propose to integrate RNA-seq and mass spectrometry based proteomics. More precisely, and knowing transcription levels do not always reflect protein concentrations,the goal of this project will be to assess how well transcriptomic can help imputing quantitative proteomics data when peptides fall below the detection limit of the instrument.

Methodology: To achieve this goal, we propose the following roadmap:
1. Exploratory analysis of paired transcriptomic and proteomic samples. Preliminary analysis of datasets using standard pipelines and assessment of correlation levels between the two sets of data. Discrepancies between RNA and protein abundances have different sources: (1) not all RNAs are translated into proteins; (2) proteins and RNA have different half-lives; (3) some proteins are transported from other cell-types.
2. Develop a novel method to estimate protein abundance using jointly transcriptomic and proteomic data. Leverage the high quality information provided by the transcriptomic data to build a new predictor of protein abundance through the transfer learning / domain adaptation frame-work.
3. Facilitate reproducible and open science by sharing the method in a high quality open-source package.

Scientific environment
1. Within the Fundamental Research division of CEA Grenoble, the lab Exploring the Dynamics of Proteomes (EDyP -http://www.edyp.fr/web/) gathers multiple scientific areas of expertise (ranging from biology to applied mathematics) with the aim to develop analytical and computational methodsthat improve the proteome coverage of complex biological samples
2. The TIMC-IMAG -https://www-timc.imag.fr/en/lab gathers scientists and clinicians towards the use of computer science and applied mathematics for understanding and controlling normal and pathological processes in biology and healthcare. Within the lab, the team BCM (Biologie Computationelle et Mathématique) focuses on developing data-driven and modeling methods for biology, living systems, and to better support our healthcare system.
3. This project will be financially supported by the artificial intelligence for high throughput biomedical investigations program of the Grenoble Multidisciplinary Institute for Artificial Intelligence (MIAI -https://miai.univ-grenoble-alpes.fr/), which fosters academic collaborations between Grenoble hospital, academic labs (among which TIMC-IMAG and EDyP), and artificial intelligence industry.

Profile: The profile sought is that of a graduate student (Master degree or equivalent) in Computer Science (Major in Artificial Intelligence, Data Science, or Bioinformatics) or in Applied Mathematics (Major in Signal Processing or Statistics) who has a strong interest in interdisciplinary work in biology. They must have programming skills (R or Python) and be fluent in either French or English. Applicants must send their CV to:
– Nelle Varoquaux, CNRS researcher, TIMC-IMAG: nelle.varoquaux@univ-grenoble-alpes.fr
– Thomas Burger, CNRS researcher, EDyP-lab: thomas.burger@cea.fr

More information here : http://pagesperso.litislab.fr/~bgauzere/phd_gan.pdf

Ces graphes, complexes, peuvent être unipartis ou bipartis, évoluer au cours du temps, s’organiser selon plusieurs couches, etc. La question de l’évaluation s’articule en général autour de la satisfaction de l’utilisateur·ice, quantifiée par une mesure de précision. Pourtant, c’est loin d’être le seul critère intéressant [2]. En particulier, le risque d’un système de recommandation trop proche du profil de l’utilisateur·ice est de l’enfermer autour d’une thématique donnée (bulle de filtre [3, 1]). La question de la diversité des recommandations d’un système est à creuser, tant à l’échelle individuelle (pour diversifier les recommandations d’une personne) que collective (pour couvrir l’ensemble des éléments du système). Par ailleurs, la possibilité d’expliquer les recommandations faites est cruciale dans de nombreux contextes. Il existe aujourd’hui une littérature large sur la diversité dans les systèmes de recommandation en général, mais celle-ci n’aborde que marginalement la question des graphes, alors même que leur versatilité n’est plus à prouver dans le domaine de l’analyse de données. Plusieurs travaux ont déjà démontré la faisabilité de cette approche [4], et il s’agit à présent d’aller plus loin.
Le but de ce stage est donc d’explorer la question de la diversité dans les systèmes de recommandation sociaux : est-ce qu’une telle modélisation, en permettant la visualisation immédiate d’un voisinage (et donc des contributions aux recommandations), est utilisable pour proposer des recommandations individuellement plus diverses? Qu’est-ce qui, dans la structure de graphe, permet de comprendre les recommandations effectuées? Comment peut-on (et veut-on) quantifier cette explicabilité?

Offre complète : https://tiphaineviard.com/files/Stage_M2_RFS_TV.pdf

Cordialement,

Gregory

(english version)
Dear all,

in SPARTA project (sparta.eu), we are offering a postdoc position (F/M) for one year
on the topic of intrusion detection systems and security information and event management systems assessment.
The candidate will develop an approach enabling the assessment of the robustness of such systems through:
– the definition of appropriate metrics
– the design of an assessment methodology including pratical traffic injection tests, generated automatically using generative and adversarial models
– the design of a multi-objective optimisation model

We are looking for the following profile:
– EITHER a Ph.D in machine learning (generative learning, adversarial learning, reliability of ML) with competences in cybersecurity
– OR a Ph.D in cybersecurity (intrusion detection, alert correlation) with competences in machine learning

More details: https://www.telecom-sudparis.eu/en/wp-content/uploads/sites/3/2020/02/FP-post-doctorant_RST-anglais.pdf

Best regards,

Gregory

Un poste de Maître de conférences 61/27 : https://www.galaxie.enseignementsup-recherche.gouv.fr/ensup/ListesPostesPublies/ANTEE/2020_1/0951376E/FOPC_0951376E_4055.pdf

Laboratoire de rattachement : ETIS, UMR CNRS 8051 (www-etis.ensea.fr).
Se référer aux profils des postes pour les contacts enseignement et recherche.

Le post-doc proposé intervient dans le cadre d’un transfert technologique pour
une entreprise développant un assistant culinaire. Différents datasets de
recettes de cuisine ont récemment été rendus disponibles et permettent
l’apprentissage de réseaux de neurones profonds (voir par exemple [1]). Dans le
cadre de ce projet, on s’intéresse à proposer des aménagement de recettes de
cuisines suivant les ingrédients disponibles (réparation de recettes).
Le post-doctorant aura pour tâche
* Organiser différents jeux de données à des fin d’apprentissage
* Expérimenter, évaluer et combiner les architectures les plus pertinentes pour
un système de recommandation culinaire.
* Une ou plusieurs publications des travaux réalisés

Profil recherché pour le poste :
* Doctorat en Informatique
* Spécialisé dans le domaine de Deep Learning
* Familiarisé avec les frameworks les plus courants (keras, pytorch, tensorflow,
etc.)

Pour plus de détails : http://cazencott.info/dotclear/public/offers/2019-04-Postdoc-ANR.pdf

Centre for Computational Biology (CBIO):
The Centre for Computational Biology is part of Mines ParisTech, one of the major engineering schools for applied mathematics in France. Our research is dedicated to the development of methods and tools in the field of Machine Learning, Statistics and Computer Vision in order to analyze massive data generated in life sciences and medicine. We work on a broad range of applications, from questions in fundamental life science to precision medicine. The Centre for Computational Biology has a partnership with INSERM, the French national health institute, and the Institut Curie, a major hospital and research center dedicated to cancer. This partnership provides access to the infrastructure and facilities of the Institut Curie and facilitates collaborative projects with other groups at the Institut Curie, as well as data sharing. Our laboratory is located in the heart of Paris and we therefore benefit from an exceptional scientific and cultural environment.

Position summary:
We are seeking a senior scientist with several years of experience as an independent researcher and with an excellent track record in the fields of Machine Learning and Statistics applied to biology and/or medicine. In particular, we would like to consolidate our activity in the fields of genomics and analysis of next-generation sequencing data with applications both in fundamental biology and medicine. Any previous experience in these fields would be an advantage. The successful candidate is expected to work in a team of 4 permanent researchers on exciting questions in computational biology, to develop his/her own line of research for which (s)he will secure funding, to supervise PhD students and Postdocs and to contribute to the scientific animation in his field of research at the CBIO and the Institut Curie.

Application file:
The application should consist of the following documents:
– a cover letter;
– a detailed CV;
– a research project (5 pages limit);
– a list of publications and of oral communications at scientific conferences;
– three recommendation letters to be sent directly to the CBIO. Alternatively, the candidate can provide contact details for three referees to be contacted directly for recommendation;
– Copies of official transcripts for all degrees at all institutions attended.
Please send the documents before 06/06/2019, to Thomas Walter, by email (Thomas.walter@mines-paristech.fr), or by mail (CBIO, Mines ParisTech, 60 Boulevard Saint Michel, 75006 Paris).

Profil recherché : titulaire d’une thèse idéalement dans le domaine du Deep-Learning pour la Robotique ou le Véhicule Autonome (ou sinon, soit thèse en Deep-Learning appliqué à un autre domaine, soit thèse dans le domaine Perception ou Planification en Robotique ou Véhicule Autonome), et idéalement au moins une 1ere expérience post-thèse dans ces mêmes domaines [mais tenure-track juste après thèse aussi envisageable].

Pour + de détails sur le laboratoire, la fiche de poste et les modalités de candidature, voir la page ci-dessous :
http://caor-mines-paristech.fr/fr/2019/03/chargee-de-recherche-en-intelligence-artificielle-pour-la-robotique-et-les-vehicules-autonomes/

Description of the project and postdoctoral missions:

Social networks are regularly used to express opinions on public and political events or to disseminate opinions on sensitive topics (hate speech, hooliganism, racism and nationalism, etc.). The objective of the SAPhIRS project is to study opinion propagation within social networks: to identify the key mechanisms for disseminating information and opinion and to identify leaders of influence. Particularly in Security field, we will focus on Twitter detection and analysis of hamessages calling for hatred or violence, monitoring their spread and detection of influential actors.

As part of this project, we propose a 12-month postdoctoral position in machine learning for opinion mining, sentiment analysis and the detection of changes of opinion in Tweets. For this purpose, we plan to use state-of-the-art methods in NLP based on deep-learning neural networks, and especially recurrent neural networks with internal memory such as LSTMs or GRUs.

In other words, the main tasks would be:

To annotate tweets automatically according to an opinion: supervised classification problem;
To automatically identify messages containing the expression of radical ideas, in English, in French and in Arabic chat alphabet (transliteration of Arabic in Latin alphabet, also called arabizi or arabish): problem of supervised learning on unbalanced classes and possibly weakly supervised learning;
To detect changes of opinion in user Tweets sequences: detection of anomalies and breaks in a time series.

The main difficulties come from the encoding of input data (short texts from Twitter, in French and Arabizi) for which language models remain to be defined, and in the design and learning of adapted recurrent models dedicated to these three tasks.

Profiles:
Candidates must have a doctorate in Machine Learning with, if possible, experience in NLP and/or Deep Learning. Knowledge of recurrent networks and Arabizi would also be a plus.

Contractual conditions:
The contract will be 12 months starting as soon as possible, with a gross salary of approximately 3500 €. The recruited person will work in the LITIS lab at INSA Rouen Normandy in Saint-Etienne-du-Rouvray.

Application: CV, motivation letter, recommendation letters.