Better qualifying uncertainty when studying complex phenomena

When and where will the next high-magnitude earthquake hit ? What will be the outcome of a cancer treatment ? In the field of risk assessment, some predictions or estimations can have a huge impact on people’s lives and the environment and being able to quantify their level of uncertainty is of paramount importance for decision-making. The difficulty is that some systems are so complex and the data available so abundant and heterogeneous, that carrying out the calculations associated with these tasks is extremely challenging, if not impossible. Using a probabilistic approach, Prof. Maurizio Filippone from EURECOM aims to develop methodologies that will tackle these computational issues while enabling highly accurate quantification of uncertainty. The AXA Chair on New Computational Approaches to Risk Modeling will give him every advantage to reach his objective.

« Thanks to advancements in sensor technology, it is now possible to collect and store weather information at a large scale, sequence the complete assembly of DNA, and obtain full 3D scans of the brain at millimeter-scale resolution », explains Prof. Maurizio Filippone. « Such advances make it possible to obtain an enormous amount of data that holds the key to unlock our understanding of the mechanisms involved in the characterization of risk in several disciplines ». Despite computers are becoming increasingly more powerful, this abundance of information remains extremely hard to process. « The key idea behind the project is that it is possible to carry out accurate quantification of uncertainty relying exclusively on approximate, and therefore cheaper, computations », explains Prof. Maurizio Filippone.

Using approximations to get precise estimations faster

To reduce computing time, the project will develop novel scalable approaches for quantification of uncertainty that will be able to exploit parallel and distributed computing. Indeed, EURECOM has a lot of expertise in large-scale distributed computing systems. The first step of the project will consist of developing and implementing the mathematical, computational and statistical methodologies necessary for this new breed of computational models. The second will be to prove the project’s capacity to accurately quantify uncertainty leading to a better modeling of risk in several domains compared to current methods.
These approaches can have major implications in the development of medical treatment strategies or environmental policies, for example. Better calculation of uncertainty could provide invaluable help for decision-makers faced with difficult decision-making tasks. Following underwater seismic activity, should nearby coastal cities be evacuated or not ? Is it worth giving a patient a specific treatment given his medical record ? « The main focus of my research is on life and environmental risk, but technically the same methodologies could be applied to the fields of finance or even social sciences », Prof. Filippone points out.