Fujitsu, in partnership with the Singapore Management University (SMU) and the Agency for Science, Technology and Research (A*STAR)’s Institute of High Performance Computing (IHPC), has launched the Digital Platform Experimentation Project. This new initiative forms part of the joint research and development activities at the Singapore Urban Computing and Engineering Centre of Excellence (UCE CoE).
These are supported by the National Research Foundation Singapore (NRF) under its UCE Corporate Lab@SMU, one of the corporate laboratories that it has set up to encourage public-private research and development partnerships between universities and companies.
The project will nurture local talent and capabilities at the intersection of artificial intelligence, deep learning, and quantum-inspired computing. A*STAR’s IHPC will play a key role in the deep learning and related AI capabilities for the project. SMU’s School of Information Systems will strengthen the quantum-inspired computing and related AI optimisation capabilities.
Combining deep learning, AI, and quantum-inspired computing technologies into a single computational service-delivery platform will help solve very complex, large-scale, real-world problems —particularly combinatorial optimisation problems at the core of planning and scheduling scenarios.
The project also taps on the CoE’s research and development capabilities to implement Fujitsu technologies and accelerate the development of commercial applications using high performance optimisation. This will further establish Fujitsu’s global quantum-inspired and AI eco-system, with the SMU installation marking the first-in-the-world, on-premises deployment of the Fujitsu Digital Annealer platform.
As the project progresses, Fujitsu, A*STAR, and SMU will work with key institutions and stakeholders in Singapore’s quantum computing community.
Dr Lim Keng Hui, Executive Director of IHPC, said, “We expect to see quantum-inspired computing exceed the limits of conventional computing in this modern age of digitalization. Through our collaboration with Fujitsu and SMU, A*STAR will develop algorithms and methodologies for resource-efficient machine learning. This will reduce memory footprint, complexity and demonstrate real world use cases for industry applications. In the longer term, we aim to deploy these technologies to address complex challenges faced in experimental and computational science.”
The Digital Annealer
According to Rio Yamaura, VP of Fujitsu’s New Solution Business Division, the Digital Annealer works with traditional bits, and does not require specialised cooling systems to run. It is a computing device that forms a bridge between traditional computing and quantum computing, and will pave the way for more rapid adoption of quantum computing when it is stable and commercialised in the future.
The Digital Platform Experimentation Project marks the world’s first on-premises installation of the Fujitsu Quantum-Inspired Computing Digital Annealer. The Digital Annealer is claimed to provide an alternative to quantum computing technology, which is at present both very expensive and difficult to run. Using a digital circuit design ‘inspired’ by quantum phenomena, the Digital Annealer focuses on rapidly solving complex combinatorial optimization problems without the added complications and costs typically associated with quantum computing methods. The Digital Annealer will play an important role in this initiative by allowing the partners to explore novel problem-solving approaches and methodologies for a wide variety of potential real-world applications. Use cases to date include portfolio optimization, drug discovery, factory optimization, inventory management, and digital marketing.
Another important aspect of the Digital Platform Experimentation Project is the demonstration of machine learning technology through Fujitsu’s Digital Transformation (DX) Services and Platforms, which include technologies that accelerate deep learning for new applications and solutions in a variety of industries. These deep learning capabilities will prove increasingly important with the growth of edge computing and IoT devices.
In this project, SMU will benchmark Digital Annealer with exact commercial solvers (such as CPLEX and Gurobi) as well as other heuristic methods to solve complex combinatorial optimization problems. Classical methods will be combined with quantum-inspired methods to discover new hybrid algorithms that run on conventional computers to tackle practical use cases in resource planning and scheduling, such as designing daily schedules for ambulance and police cars to respond to crimes and emergencies in a congested city. This research will help to optimize resources toward a smart, safe and sustainable city.
A*STAR’s IHPC will contribute capabilities in developing deep learning models on real-life use cases with video data analysis for security applications, such as video anomaly detection, video action classification and real-time crowd analysis. The research focus aims to shorten video training time and significantly reduce memory footprint requirements.