Post-doc position on generative models for inverse problems and simulation-based inference (currently open).

We have an opening for a full-time post-doc position. The research target is to develop machine learning methods to solve inverse problems accounting for the inherent uncertainty of the inverse problem. We will formulate these problems as inference problems with prior scientific knowledge available in the form of simulators. We will follow the simulation-based inference paradighm to model the posterior distribution of the simulators parameters that could have generated the observed data. Of partricular interest are the settings in which the domain knowledge is incomplete and/or imperfect. Classical approaches to simulation-based inference, e.g. Approximate Bayesian Computation, suffer from high complexity costs in particular in high dimensional spaces and machine learning has great potential to address such problems [1]. We will build upon our recent work on generative learning approaches to solve inverse problems using forward models and simulation [2,3]. We will explore ways to explicit encode, and complete when necessary, available domain knowledge, resulting to what is known as grey-box models. Such models respect by construction the domain knowledge, will not produce implausible outputs, and have strong extrapolation performance.

The position is funded by a Swiss National Science Foundation interedisciplinaty project, with funding secured for four years. The project brings together teams from geology, seismology and machine learning with the overall objective to streamline the passive seismic exploration by developing new, machine-learning based, analysis tools in order to identify potential targets for geothermal exploitation. Within this context the DMML team will develop generative models to invert the sheer wave velocities measured in earth's surface in order to recover velocity models of the subsurface.

We seek strongly motivated candidates dedicated to high quality research. Candidates should have (or be close to obtaining) a Phd in machine learning, ideally in the area of generative modelling, and a strong research track-record attested by high quality publications in relevant machine learning venues such as ICML, NeurIPS, ICLR, AI-STATS, UAI, etc. The selected candidate is expected to demonstrate a high degree of independence and autonomy, drive their own research and actively contribute to the scientific development of the group through their knowledge and expertise as well as by proposing and contributing to group activities such as readings, schools, workshops etc. They are expected to participate in the supervision of a PhD student that will also be working within the same project. There is also a possibility to participate to teaching activities.

The DMML group consists of around ten researchers at the PhD and Post-doc level, working in different areas of machine learning, such as generative models, imitation learning. The team collaborates closely with the VIPER group,, from the computer science department of the University of Geneva headed by Prof. Stephane Marchand-Maillet. In addition the group is involved in a number of national and international research projects. We offer ample opportunities and support for scientific development, e.g. providing funding for conferences, schools, research visits and exchanges etc. We strive to provide a research environment in which researchers can focus on their research and allow for space and time to develop solid ideas.

If interested, please send the following to

  • academic CV (max 2 pages)
  • pointers to their two most important publications
  • one page motivation letter explaining why the candidate is suitable for the position and what they can bring to the DMML group
  • 1000 word research proposal on the research topic of generative models for discrete data structures
  • contact details of three referees (do not send reference letters)
  • copies of diplomas (PhD, MSc, BSc) and academic transcripts

Deadline for applications: 30/07/2022.

In case of any further questions, please contact

[1] K. Cranmer, J. Brehmer, and G. Louppe. The frontier of simulation-based inference. Proceedings of the National Academy of Sciences, page 201912789, 2020.
[2] N. Takeishi and A. Kalousis. Variational autoencoder with differentiable physics engine for human gait analysis and synthesis. In NeurIPS workshop on Deep Generative Models applications, 2021a.
[3] N. Takeishi and A. Kalousis. Physics-integrated variational autoencoders for robust and interpretable generative modeling. In Advances in Neural Information Processing Systems 34, 2021b.