Leading CNRS project on hybrid AI launched in Singapore

A vast collaborative project on hybrid artificial intelligence, led by the CNRS's subsidiary in Singapore and allocated a budget of €35 million over five years, will begin in late 2021.

DesCartes – behind this very French name is one of the CNRS’s largest collaborative projects, the Programme on Intelligent Modelling for Decision-making in Critical Urban Systems, which is dedicated to hybrid artificial intelligence. With a consolidated budget of €35 million over five years, this programme was recently selected by the government of Singapore, and will be based at its international research hub, the CREATE campus.1 The CNRS and 12 universities and Grandes Écoles2 are involved on the French side, together with five Singaporean universities3 and the country’s A*STAR institute.4 “This ambitious programme will enable us to set our priorities. It is not a response to a call for projects, as it was the CNRS that made a proposal to the National Research Foundation (NRF),”5 points out Dominique Baillargeat, Director of the very first CNRS foreign subsidiary, CNRS @ CREATE, which has been based at the CREATE campus in Singapore since 2019.

AI: France – Singapore, a shared ambition

The primary objective of the DesCartes programme, initiated in January 2020 under the management of Francisco Chinesta,6 is decision-making in “critical urban systems”, one of the many research challenges related to the “smart city” concept. “Singapore's CREATE hub did not have a project in this field. With DesCartes, the CNRS has secured a prominent position within the campus, while responding to the needs identified by Singapore for implementing its smart city policy by 2030, via its national AI strategy launched in 2019, and led in particular by the NRF's AISG programme,” Baillargeat explains. DesCartes also makes it possible to participate in the French national artificial intelligence plan, as four Interdisciplinary Institutes of Artificial Intelligence (3IA) have been created, and could potentially take part. “This is one way to extend the research conducted at these institutes, and to expand one of its components to include Singapore. The CNRS has made AI and its applications one of its priorities, especially in connection with international partnerships.” The broader context is also largely favourable, as the two countries recently established a strong AI research collaboration, concentrating on four specific areas: data and its applications; verifiable and explainable AI; natural language processing; and AI and human-machine interaction.

“This is about combining the strong expertise of both countries,” Baillargeat adds. The Singaporean consortium is renowned worldwide in the field of smart cities and disciplines central to AI. The French consortium, on the other hand, enjoys international recognition in AI hybridisation technology and engineering, as well as in the human and social sciences, which will be decisive in placing citizens at the heart of the programme. 

From optimising the electricity network to managing traffic for delivery drones

DesCartes will develop hybrid AI methods that combine AI with models based on knowledge (physics, engineering) in order to ensure reliable real-time decision-making that “focuses on individuals” and is “in step with society”. This will all be in response to complex situations related to critical urban systems in the context of Singapore's “Smart Nation” initiative. This research will, for example, be applied to urban mobility and energy management, or to anticipate the needs of future transportation networks, intelligent industries, and smart buildings. “For instance, we intend to develop solutions that optimise supply with demand in the electricity network or relieve traffic for drone taxis or delivery drones, in addition to developing smart predictive maintenance applied to industrial tools,” Baillargeat points out. This includes urban systems that will be capable of analysing data on the population's use of services in order to make optimisation decisions. These tools will be developed by some 30 PhD students, 50 postdoctoral fellows, and more than 80 academics and researchers from Singapore and France, some of whom will be based in the City-State for long-term stays financed by the programme. 

“Singapore is a genuine showcase for innovation, one that is open to emerging technologies. It is the ideal place for demonstrating that hybrid AI is a groundbreaking technology suitable for the critical infrastructure of smart cities.” The ambition of DesCartes is to deploy this technology on industrial-scale case studies by involving governmental agencies and local industries. “The applications, of course, will be global.”

This article was first published on June 24 by CNRS.