Fujitsu and Ben-Gurion University Embark On Joint Research At New Center In Israel For Precise And Secure AI

By Ron Chen, Correspondent ASEAN Region and Ifeoma Asika, Correspondent Campuses

Fujitsu Limited and BGN Technologies, the technology transfer company of Ben-Gurion University of the Negev (BGU), have signed a three-year comprehensive joint research agreement to develop technologies and solutions to contribute to the realization of safe, real-world applications of AI and machine learning technologies.

The newly-established “Fujitsu Cybersecurity Center of Excellence in Israel” (Fujitsu CCoE IL) hosts a team of approximately 20 researchers including Prof. Yuval Elovici and Prof. Asaf Shabtai. The lab is headed by Prof. Elovici, Department of Information Systems Engineering and Head of BGU’s Cyber Security Research Center located at the Cyber@BGU labs in the advanced technology park at the university’s campus in Israel.

With this initiative, Fujitsu and Ben-Gurion University aim to accelerate joint research and development into technologies that offer solutions to threats posed to AI models and machine learning, improving the trustworthiness and security of AI-based systems and software that increasingly impact our daily lives.

Chief Technology Officer of Fujitsu Limited Vivek Mahajan commented, “We are delighted to collaborate with Ben-Gurion University, which boasts world-class software development capabilities in the security field. I’m confident that the establishment of the Fujitsu CCoE IL will yield a variety of promising new security technologies for AI, and I look forward to a fruitful partnership with Ben-Gurion University as we work in tandem to resolve some of the challenges facing our increasingly digitized society.”

Prof. Yuval Elovici, said, “We are excited to enhance our collaboration with Fujitsu, a world-leading information and communications technology company. Together we can make systems that are based on AI more robust to attacks that focus on the AI component.”

“This joint strategic research collaboration is a testament to Ben-Gurion University’s leadership in the cybersecurity space,” added CEO, BGN Technologies, Josh Peleg. “The multi-year collaboration will improve AI security and help create a safer world while strengthening the University’s positioning as a global player in the AI, machine learning and cybersecurity arenas.”

Growing Opportunities, Growing Threats

AI and machine learning technologies play a growing role in society, with promising applications emerging that will increasingly affect our daily lives, including the improvement of public safety through analysis of human behavior in surveillance video data, product quality control through the detection of abnormalities, product recommendations, as well as to improvement of autonomous driving.

With the growth and expansion of these technologies, however, the need to improve AI security represents an urgent priority. Hostile entities use increasingly sophisticated techniques to threaten critical infrastructure and systems by stealing and leaking confidential information contained in AI datasets. For instance, adding special noises to video data can cause the AI to misidentify people or misdetect actions. Fujitsu CCoE IL will address these issues through research on AI security to detect such attacks, protect against adversarial access to AI systems, and to make the models more robust against attacks.

As part of their new partnership, Fujitsu and BGU will focus on research that confronts these threats to make AI technologies safer and more trustworthy. For one of the Center’s initial research projects, researchers from Fujitsu and Ben-Gurion University will apply an AI model (Out of Distribution detection model (1)) that can detect new, unknown types of threats, such as attacks on drones and fraud in network communications and develop technologies able to appropriately cope with the rapidly evolving attack methods. The Out of Distribution detection model will additionally be applied to detect product malfunctions in cases where unexpected data appears intentionally or accidently.