Can you tell us about your research project at Inria?
I've been working at Inria for about ten years now as part of a joint team with Ecole Normale Supérieure and the French National Center for Scientific Research (CNRS) on machine learning, a sub-branch of artificial intelligence. It's about developing methods for automatic processing of very large quantities of digital data. In machine learning, we study the past to better predict the future. Using large amounts of data, we try to deduce rules and produce a predictive system. With machine reduction, the algorithms sometimes become slow. Our goal is to speed up the algorithms in this multiple-machine environment, and to have them make relevant predictions. We are seeking a compromise between the speed of calculation time and the accuracy of the prediction. We also want to make it possible for the algorithmic methods that we are developing to be used worldwide. That's why our code is open source, so that people will use our algorithmic methods in their field, whether it be computer science, vision, etc.
You have been awarded an ERC grant this year. For which project was it given?
The Sequoia project, which handles the speed of algorithms and the relevance of the result using numerous machines and unstructured data. We currently have very big, ill-posed problems. The goal is to create well-formulated sub-problems and resolve them.
What triggered your desire to work on this project?
I like machine learning because it's a field that interacts with many others, including statistics, mathematics, as well as language and vision. Therefore, interacting with colleagues from these disciplines is important. It's also very interesting to see the link between maths and real life. I like going back and forth between theory and application.
What does this grant mean for you?
It's a new step (editor's note: the first was awarded in 2009). It's a rather strong academic recognition that's important for a career. The ERC is especially important to obtain significant funding for our research work. Now we no longer have to worry about the financial aspect, which is a good thing. Even if the grant is awarded on a personal basis, it acknowledges the work of an entire team. I do not work in isolation. I have received a lot of assistance from close colleagues and Inria's Paris centre, who helped me in my work on the ERC application. They contributed greatly to creating a rigorous and ambitious project. Such professional support is very important.
How are you going to make use of the grant?
We are going to recruit students and post-docs to expand the current team. We need the staff to work on the Sequoia project for five years. We hope to hire seven people. We also need machines. We must work on the same machines that people use. We need to be able to test the algorithms in actual conditions and on the same architectures as the end users, or in the cloud.
Do you have other avenues of research that you want to explore in the future?
I have no strategic plan beyond five years, since it will all depend on what other researchers are doing and how things evolve in the meantime. And there is also the matter of professional encounters, of course.
Five key dates in the career of Francis Bach
- 2005: PhD in Computer Science, University of California at Berkeley
- 2007: Joins Inria.
- 2009: First ERC Starting Grant.
- 2011: Research team (approximately 20 people) created following his first ERC.
- 2012: Inria Young Researcher Award
- 2016: Second ERC Consolidator Grant.