PhD Position F/M Surrogate modeling for simulation-based inference with application to genomics data

The Inria Grenoble research center groups together almost 600 people in 23 research teams and 7 research support departments.

Staff is present on three campuses in Grenoble, in close collaboration with other research and higher education institutions (University Grenoble Alpes, CNRS, CEA, INRAE, …), but also with key economic players in the area.

Inria Grenoble is active in the fields of high-performance computing, verification and embedded systems, modeling of the environment at multiple levels, and data science and artificial intelligence. The center is a top-level scientific institute with an extensive network of international collaborations in Europe and the rest of the world.

Within the framework of a partnership (you can choose between)

• Collaboration with Pedro Rodrigues and Nelle Varoquaux within the BONSAI project (ANR JCJC).

Is regular travel foreseen for this post ? Yes, travel costs will be covered within the limits of the scale in force.

Context and Objective:

Surrogate modeling has recently received increasing attention in the context of simulation-based infer- ence to address the intractability of the likelihood [Brehmer et al., 2020]. All these approaches amortize the approximation by learning the joint dependence on the simulated data and the simulator’s parameters. While such amortization yields computational benefits, modeling the joint dependence as a smooth function is rather restrictive as it can fail to capture discontinuities arising from phenomena such as phase transitions and which are of particular interest in physics experiments [Ricci-Tersenghi et al., 2019]. In this work project, we aim to leverage the flexibility of deep learning to learn surrogate models for simulation-based inference without suffering from the limitations of amortization. We propose instead to decouple the surrogate model from the simulator’s parameter for more flexibility. Such a decoupling effectively results in a bilevel optimization problem where the upper-level objective learns the simulator’s parameters while the lower-level objective learns a surrogate model for a given parameter value. We divide the work into three tasks: task 1 introduces a framework for black-box simulators. Task 2 introduces a framework for white-box simulators, those for which additional information is available, while task 3 applies such as framework to a real-world application from genomics.

Assignments : 

The PhD project aims at developing methods based on surrogate modeling for simulation-based inference, with application to genomic data. 

Responsibilities: The recruited person will take initiatives to formulate the problem and address it in a rigorous manner.

Mains activities:

• Publish in top tier conferences and journals
• Produce high quility open-source software
• Participate to the research community (participation to conferences, workshops, etc)

Technical skills and level required : Excellent coding skills, strong mathematical background.

Languages : English

Interpersonal skills : Drive and perseverance

• Subsidized meals
• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Possibility of teleworking (90 days / year) and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Complementary health insurance under conditions

1st and 2nd year: 2 100 euros gross salary /month

3rd year: 2 190 euros gross salary / month