Post doctoral position at EPFL Switzerland
Application of kernel methods to large-scale estimation problems is often limited by the computational complexity of learning algorithms.
However, one often has a priori information about the regression problem in terms of relations with a physical process. The project will focus on machine learning techniques exploiting this knowledge for the derivation of approximate estimators that can be computed in a distributed fashion.
A related aim is to investigate the impact of machine learning approaches to the solution of Partial Differential Equations (PDEs). PDEs offer a broad and flexible framework for modelling and analysing a number of phenomena arising in fields as diverse as physics, engineering, biology, and medicine.
Many times, the simulation of PDEs is performed within an optimisation loop that may involve geometric and/or physical parameters, meaning that the PDE itself needs
to be simulated many times. This often results in a computational intensive work.
The goal is to explore machine learning approaches for modelling the dependence of PDE solutions from geometric parameters and to understand to which extend machine learning techniques can reduce the computational burden on specific problems.
Prospective Post-docs should have:
- a PhD degree from a recognised university
- a strong background in machine learning
- creativity and motivation for addressing mathematical problems
- excellent English language skills