Research Programmes

Sections
Left Column
Image
Image
008.jpg
Right Column
Text Area

The proliferation of AI applications such as autonomous vehicles, medical diagnosis, video image analytics has fuelled the transformation of many industries. With the introduction of specialised AI chips, it makes training and running AI models much more efficient and effective and provides the benefits of handling AI tasks faster and consuming less power than general-purpose chips. According to a recent Research and Markets report, the global AI chips market is forecasted to grow by US$ 54.03 billion, with a CAGR of 42% during 2020-2024.  

To cope with this emerging demand on hardware, ACCESS is putting Hong Kong on the global map of AI chip and hardware design with research focus on advancing integrated circuit (IC) design technologies to enable novel data-centric computing paradigms supporting a wide range of AI applications. We also focus on the research theme of designing customized AI chips to realise ubiquitous AI applications in society.

The research agenda in ACCESS is organised into 4 research programmes, addressing four key technical areas:

First Column
Text Area

RP1

Text Area

Enabling Technology for Emerging Computing Systems

Text Area
Second Column
Text Area

RP2

Text Area

Architecture and Heterogeneous System Integration

Text Area
Third Column
Text Area

RP3

Text Area

AI-Assisted EDA (Electronic Design Automation) for AI Hardware

Text Area
Fourth Column
Text Area

RP4

Text Area

Hardware-Accelerated AI Applications

Text Area
Text Area

Within five years, ACCESS research will lead to AI hardware with at least 1000× improvement in performance/ energy over the conventional state-of-the-art hardware. Even more exciting, the research will result in design methodology capable of taming the design complexity and shortening the design cycle to better meet the time-to-market requirements, and enabling large-scale on-chip artificial neural networks capable of rivaling biological counterparts in performance and energy efficiency in intelligent information processing tasks.