EPFL launches Center for Intelligent Systems

EPFL’s new Center for Intelligent Systems (CIS) will act as a research promotion platform for bringing together experts in machine learning, data science, computer vision, cyberphysical systems, and robotics.

Initially conceived to unite the schools of Engineering (STI), Computer and Communication Sciences (IC) and Basic Sciences (SB), CIS’s unique mission is to connect and support all EPFL researchers working in fields related to intelligent systems. These fields are developing technologies that, when brought together, can be used to construct intelligent systems capable of making complex, nuanced decisions in challenging, dynamic environments.

CIS Executive Director Jan Kerschgens says his goal for the center is to get researchers out of their research silos, and their comfort zones, to pursue ambitious and collaborative projects.

“We have never had three schools come together like this to promote research and technology transfer under the umbrella of intelligent systems. The CIS is also 100% synergistic with other EPFL centers like the Center for Digital Trust and the Swiss Data Science Center,” Kerschgens says. 

Included within the CIS is the EPFL unit of ELLIS, a European initiative aimed at bringing together experts in machine learning and intelligent systems. The center’s website also includes a list of EPFL courses for students interested in developing key competences in intelligent systems, including software and connectivity, big data management and analysis, smart computing systems, and embedded devices. Finally, CIS will act as a point of contact with industry, as well as Swiss and international stakeholders, to create long-term partnerships and accelerate innovation in complex systems.

“We are at a crossroads, moving from the intelligent design of systems towards the design of intelligent systems. These systems will contribute to science, education, industry, and the economy at large. They will enable scientists and engineers to explore complexity and the design space more fully. Through CIS, EPFL is positioning itself to lead in this domain,” says Ali H. Sayed, dean of STI.

Hitting the ground running

Since January, Kerschgens has overseen the launch of CIS’s “Get to Know your Neighbor” seminar series for the EPFL community. He is now planning, with the support of the CIS steering committee, a colloquium series that will invite distinguished speakers in intelligent systems. Other plans include a winter school and joint lab funding for postdoctoral researchers.

Although the CIS launch coincided with the onset of the coronavirus pandemic, its members are already leveraging the new resource to brainstorm collaborative research projects, notably in the areas of health and medicine.

“We want to have an impact, on COVID-19 research for example, as a benefit of having experts working together,” IC Dean James Larus told attendees at the first CIS seminar on April 6th.

Indeed, STI professor and CIS steering committee member David Atienza recently announced the development of the Coughvid app in his Embedded Systems Lab (ESL), in collaboration with students from STI, IC, and SB. The concept for the app, which aims to identify the telltale dry cough of a COVID-19 patient by using artificial intelligence to analyze sound waves, sprang from a CIS discussion on how intelligent systems could be used to help the Swiss healthcare system and, in this particular case, mitigate the pandemic’s effects. The app recently won one of the top prizes at the Swiss hackathon, LauzHack.

Atienza says that the team is currently working with the Lausanne University Hospital (CHUV), Unisanté, and the University Hospital of Bern (Inselspital) to medically validate the Coughvid protocol. He points out that this collaboration is a perfect example of how CIS can motivate external experts to work with EPFL researchers: “EPFL can benefit from a center like this whenever you have complex, large-scale problems like the COVID-19 pandemic: because we are already working together within the framework of CIS, it’s easier for collaborators to come to us with a problem to solve.”

This article was first published on 11 May by EPFL.