We envision the next-generation AI computing hub to empower stakeholders to participate in resource sharing through a high-performance, privacy-friendly computing platform.

We are building an open platform that provides state-of-the-art algorithm paradigms and high performance computing resources for researchers in the cross-disciplinary of machine learning, data science, civil and environmental engineering and public policy. The platform will power smart city applications, starting with Hong Kong.

We are building a scalable AI infrastructure that is designed for the evolving machine learning researches, including Graph Neural Network and Federated Learning:

· High Performance Cluster with CPU/GPU/FPGA
· Privacy-preserving AI & Acceleration Technology
· Serverless Computing Cluster Management
· Reliable Massive Data Storage

CityNet aims to promote a smarter Hong Kong by fueling smart city applications with heterogeneous data from different sources.

CityNet consists of multi-modality data collected from Hong Kong government, including POI (point of interest), CCTV camera recordings and images on streets, transportation status for different methods (bus, taxi, car) and weather information. Machine learning models can utilize these streaming graphs from different aspects together to mine more interesting patterns and make better predictions.

From smart travel guides to taxi scheduling for better traffic utilization, CityNet will be the key to enabling various important smart city applications, including smart economy, smart mobility and smart living.

Take smart mobility as an example, based on the computation power of our TACC platform, we will implement an AI-enhanced taxi dispatching system. With the Hong Kong local data from the Transport Department and HKO, our framework will effectively improve the transportation system capacity in this busy city.