Post-doctoral fellow in Robust and Explainable AI

Job description

TELECOM PARIS

Télécom Paris, a founding member of Institut Polytechnique de Paris and a part of IMT (Institut Mines-Télécom), is ranked as one of the top 5 engineering schools in France. Recognized for its invaluable corporate connections, Telecom Paris is an institution with a strong sense of community and international representation. This public higher-education establishment guarantees excellent employability in all sectors and is the leading engineering school for the entire digital spectrum (from hardware to software technology).

With its high-level instruction and innovative pedagogy, Télécom Paris is at the heart of a unique innovation ecosystem which is based on collaborative classroom engagement, an emphasis on project-based training, and multidisciplinary research. Our scientific faculty are affiliated with two key research laboratories; first, the LTCI laboratory, which is recognized by the HCERES as a leading facility in the field of digital sciences and has a remarkable international outreach. And second, the i3 laboratory, Interdisciplinary Institute of Innovation (I3 - UMR 9217 of the CNRS), which carries out multidisciplinary research focused on innovation and in collaboration with the École Polytechnique and Mines ParisTech.

Based in Palaiseau, France at the heart of the Institut Polytechnique campus alongside the École Polytechnique, ENSTA, Télécom Sud Paris and ENSAE, Télécom Paris also has a central-Paris incubator at the epicenter of the French start-up ecosystem.

Télécom Paris is a living laboratory for all the major technological and societal challenges including Artificial Intelligence, Quantum Computing, IoT, Cybersecurity, large-scale digital equipment (Cloud Computing), 5G/6G, and Green IT.


SCIENTIFIC CONTEXT

A postdoc position is available at the Information, Data, Signal department of Télécom Paris, French grande école inside Institut Polytechnique de Paris. This postdoc position will benefit from the effort of the group in this domain, working with two PhD students (Jayneel Parekh and Mélanie Gornet), supervisors Florence d’Alché and Pavlo Mozharovskyi, and collaborations with Winston Maxwell (Télécom Paris, Operational AI Ethics), Stephan Clémençon (DSAIDIS) and IDEMIA through the ANR project LIMPID.

The group is currently working on a novel framework named FLINT for Learning with Interpretability (see the recent article) that allows to tackle by-design learning and that can be employed for post-hoc explanation. It consists in learning jointly a pair of models, one devoted to the interpretation of the other.

MAIN RESPONSIBILITIES AND DUTIES


An axiomatic view of the loss and penalties is adopted, which yields the learning of a dictionary of high level features. This framework was successfully tested on image recognition and now next challenges are considered, to be addressed during the current postdoc position:

• continuing on image recognition: ensuring faithfulness, imposing knowledge-based properties (invariance by transformation for instance) to high level features and improving understandability of the high level feature functions by adding auxiliary tasks to solve;

• on further tasks like multi-label classification and structured prediction: development of the FLINT framework;

• definition of evaluation metrics for robustness, uncertainty, and interpretability




Job requirements

Required skills, experience, and knowledge:


  • In-depth theoretical and practical knowledge in the field of artificial intelligence.
  • Knowledge of programming language Python and frameworks for artificial intelligence, e.g., PyTorch/TensorFlow.
  • Good command of English


Preferred skills, experience, and knowledge:


  • Experience in learning of neural networks on large scale.
  • Experience in application of neural networks to real-worlds tasks.


Other abilities and skills:

  • Ability to work in a team, discussion, and scientific text editing.
  • pedagogical skills


Candidates with the following qualifications may apply:

- PhD or equivalent


APPLICATION INSTRUCTIONS

Applications should include:

- A detailed CV

- A cover letter

- Any element considered useful for the examination of the application


In order to be considered, applications must be received no later than: 01/10/2022

Scientific contact person: Florence D’Alché-Buc (florence.dalche@telecom-paris.fr) and Pavlo Mozharovskyi (pavlo.mozharovskyi@telecom-paris.fr)
Administrative contact person: Marion Taldir – marion.taldir@telecom-paris.fr


Telecom Paris is an equal-opportunity employer.
All our positions are open to individuals with disabilities.