It is said to be difficult to successfully achieve open innovation in Japan. France, where the academic sector makes a particularly large contribution to creating new businesses, is a successful model of open innovation. In this article, top French researchers talk about important factors for promoting open innovation through industry-academia co-creation.
[Fujitsu Forum 2018 Front-Line Session Report]
Development of New Businesses by French Academic Institutions
Bertrand Braunschweig, Director of the Inria (French National Institute for Research in Computer Science and Automation) Saclay Research Center, introduced trends in AI in France and Inria's activities by citing actual case studies.
France's National Strategy Focuses on the AI Field
Historically, France has been very strong in the field of mathematics. In the past, top mathematicians from France include Évariste Galois, who developed the Galois Theory, and Pierre de Fermat, who developed Fermat's Last Theorem. Currently, many French researchers are actively involved in the AI field, and a number of researchers are working at Google, Facebook, and other tech companies.
Under the government's leadership, France has been focusing on the AI field. At the end of March 2018, French President Emmanuel Macron announced a national strategy for the AI field. This announcement is based on the Villani Report on AI strategy published by Cédric Villani, a renowned French mathematician, Fields Medal winner, and Member of Parliament.
Under this AI strategy, France will invest a total of 1.5 billion euros into the AI field in the period leading up to 2022. It was also revealed that Inria will take a leading role in the research component of this future programme.
Tech Transfer: Key to Open Innovation
Inria, one of the top-ranked research institutions in France, has created many startups. Inria has eight research centers in France, including Inria Saclay, one of the two research centers in the Paris region; six of the research centers have satellite offices.
At present, Inria has researchers from more than 100 countries engaged in research and development. Inria's 2,400 employees are organized into 180 project teams. 50 Inria researchers have received grant funding from the ERC (European Research Council), which supports excellent scientists in Europe. Inria is also promoting partnerships with research and academic institutions, and 85% all 180 projects are joint with other institutions, among which the with the CNRS (French National Center for Scientific Research).
Dr. Braunschweig emphasized Inria's advantage: "Technology transfer is a factor that differentiates Inria from other academic institutions. There are many technology transfer mechanisms at Inria by which technologies and expertise are transferred to private companies. Our advantage is that we collaborate with global large enterprises, SMEs through joint innovation laboratories, regardless of their sizes."
In fact, Inria has been working on collaborative research projects with domestic companies (e.g., EDF, SAFRAN, and TOTAL) as well as global companies (e.g., Fujitsu, Microsoft, Samsung, and Facebook). "Inria is scaling up its activities by expanding our global partnerships. Collaboration with Fujitsu is part of this initiative."
Dr. Braunschweig explained, "We are particularly focused on creating startups. Inria also serves as a facilitator to follow up and support startups in building experience and fundraising. We are working to promote personnel exchanges by establishing a network that includes laboratories, large companies, and ventures." Inria has incubated more than 140 startups so far, and 75% have merged with large companies. It has also helped create job opportunities for more than 3,000 people.
Collaboration between Fujitsu, the French Government, and Inria
Next, Axel Mery, Chief Technical Officer (CTO) of Fujitsu France, introduced Fujitsu's AI business in France.
In March 2017, Fujitsu announced a five-year investment initiative of more than 50 million euros to support digital innovation in France. The goal of the initiative, which is based on collaboration with the French government, is to develop new services and acquire novel technologies. Also, in collaboration with Inria, Fujitsu has established a Center of Excellence (CoE), which is a group of experts that focus on AI.
Mery stated, "CoE focuses more on business development than research. Members strive to solve issues and meet needs by developing solutions and proposing projects. CoE also welcomes talented students and interns from Ecole Polytechnique, one of the most prestigious educational establishments in France, who are offered the chance to participate in research based on real-world business deals. This actively contributes to the fostering of digital human resources in France." During its first year of activity, CoE realized more than 10 use cases in the fields of insurance, retail, and elsewhere.
Fujitsu Co-creation Projects Realized by Industry-Academia-Government Collaboration
Fujitsu's Hirotaka Hara gave examples of case studies in the fields of manufacturing, healthcare, and social infrastructure to illustrate Fujitsu's industry-academia-government collaboration projects related to AI in Japan.
The first case study was a joint project with RIKEN (the Institute of Physical and Chemical Research). In April 2017, Fujitsu and RIKEN established the RIKEN AIP-FUJITSU Collaboration Center. Hara said, "Fujitsu is working on joint R&D centered on leading Japanese AI researchers with the aim of creating 'AI that predicts the unpredictable.' Such AI is expected to support better decision-making for people based on appropriate predictions of the future―even in the face of environmental uncertainties."
As an example, Hara introduced a project to verify AI's utility in manufacturing. The project aims to use AI to improve efficiency when developing solid electrolytes for lithium-ion rechargeable batteries. Development efficiency has been improved by employing the Bayesian inference method, which requires less computation data for simulations that previously took an enormous amount of time to calculate. The Bayesian inference method enables one to probabilistically infer the phenomena that one wants to investigate based on observed realities.
Next, Hara introduced a case study on co-creation in the healthcare field with Kyoto University. Fujitsu and Kyoto University are jointly working to apply AI to efficiently detect glomeruli in kidney biopsy images. Specifically, this co-creation project aims to develop object detection technology to support visual detection by doctors. Conventionally, identifying specific objects in a diagnostic image requires tens of thousands of images with correct data. Using the technology developed in this project, large volumes of correct data in which the positions of objects have been accurately estimated can be generated without correct data. Also, detection accuracy has been improved by more than double. Hara explained, "This project is being carried out jointly with doctors. In AI application projects, we cannot produce any results without collaborating with experts in the relevant field of specialty."
Then, Hara presented a case study on co-creation between Fujitsu and Kyushu University as an application of AI to social infrastructure. Through this project, both parties aim to develop mathematical AI techniques that are fair and acceptable to society. He also showcased a congestion prediction system at Fukuoka Airport, a matching algorithm for recommending candidate relocation destinations in Itoshima City, and a matching technology for allocating children to daycare centers in Saitama City: "These are examples of co-creation projects among academic institutions, companies, and sites in the field."
Topological Data Analysis (TDA): The Focus of Co-Creation
Fujitsu, the French government, and Inria have launched collaborative research that uses "topological data analysis" (TDA). Together, Fujitsu and the French government have been promoting co-creation activities primarily regarding AI that place the primary focus on joint research on TDA. Frederic Chazal, Inria's TDA research leader, introduced details of the study on TDA at Inria.
TDA is technology that efficiently analyzes sensor data from IoT (Internet of Things) and other devices. Using state-of-the-art mathematical technologies, various phenomena can be detected automatically. For example, we can detect infrastructure deterioration from sensor data, catch repair needs before breakdown occurs using in-vehicle sensors, and monitor user behaviors via smartphone sensors.
Top French Researchers Discuss Important Factors in Industry-Academia-Government Co-creation
Lastly, top French researchers talked about key points for promoting open innovation.
Mr. Chazal: "What is important for industry-academia collaboration is to establish good partnerships; these can be maintained only if they are mutually beneficial. A common understanding is also necessary. To that end, it is important to actively interact with each other by meeting regularly."
Dr. Braunschweig: "For successful collaboration, it is important to consider AI usage from a practical point of view at the start of collaboration."
Hara concluded the presentation with these words: "Since Japan's support system for startups has not yet been fully established, we have many things to learn from France's efforts. Fujitsu wants to use our experiences and insights gained through this collaboration with Inria in future co-creation."
- Bertrand Braunschweig Director, Inria Saclay Research Center
- Frederic Chazal
Senior Research Scientist and DataShape Project Team Leader
Inria Saclay Research Center
- Axel Mery CTO, Fujitsu France
- Hirotaka Hara
Corporate Executive Officer