The centre also taps on A*STAR’s sizeable talent pool in data science and AI, to deliver solutions for public sector agencies and industry.
Our R&D and Technologies
AI3 capabilities include machine learning, data analytics, computer vision, speech and natural language technologies, robotics, and bioinformatics. These technologies will be brought together in an integrative and coordinated manner to tackle public sector and industry needs in various domains.
Leveraging existing capabilities, A*STAR is developing new focal areas such as:
Human-Centric AI
Current day AI is primarily machine-centric and much development is required to advance AI capabilities to a level that can deeply understand humans, reason for humans, and learn like humans. By connecting AI with people at multiple levels, the objective is for AI to behave safely and support goal-oriented human-social behaviours and activities. Developments in this area can bring about more seamless, intuitive human-robot, human-machine interactions in homes and at workplaces. It will also be the enabler of many applications such as collaborative robots in manufacturing, assistive robots for rehabilitation, and personalised skills training for workers, which benefit individuals, as well as the economy and society.
Next-Generation Deep Learning
Deep learning has been a key contributor in the growth of AI and huge advances in deep learning will continue to be a force-multiplier in propelling AI capabilities forward. A*STAR’s focus on next-generation deep learning will include development of algorithms which will make technological leaps towards learning with fewer labelled samples, compression of neural networks, incorporating knowledge graphs, and white-box deep learning. Advances in next-generation deep learning will further expand the performance of AI in applications such as real-time video, audio and image processing for autonomous systems, heterogeneous analytics for predictive maintenance, machine learning on embedded and mobile devices, and security analytics for IoT.