Creating AI systems and machines that can learn to predict human intentions is the driving force behind the work of Scientia Associate Professor Lina Yao, from the University of New South Wales (UNSW) School of Computer Science and Engineering.
Yao recently took out the Cyber Security category at the Women in AI Awards.
As the director of UNSW’s Data Dynamics Lab (D2 Lab), Yao’s research focuses on enabling novel ways of human-machine interactions. She does this by using data mining, machine learning and deep learning algorithms to discover unknown patterns and develop actionable knowledge from massive heterogeneous, multi-modal, dynamic and sparse data.
“Our long-term goal is to build up the collaborative intelligence between humans and AI systems to improve human welfare and wellbeing,” she said.
“Currently, machines may be good at detecting the intentions of a human only after some observable clues are presented, or responding to cases they are experienced or trained for.
“However, responses can be somewhat robotic when a new case is presented, or the conditions or features are partially presented or observed by the AI agents.”
Machines should be able to capture data from sensors expanding far beyond human perceptions, Yao said, allowing them to adapt quickly to different environments.
Novel machine learning
Research by the D2 Lab is already contributing to AI applications in health, transportation, defence, fintech and ecommerce.
“In cybersecurity, we are working on using machine learning and AI solutions to automatically detect misinformation, such as fake reviews on shopping websites … or clickbait prevalent on the web and social media,” Yao said.
“The massive spread of misinformation through social networks has become a global risk, implicitly influencing public opinion and threatening social and political development.”
Yao and her team are also involved in a collaborative project aimed at building resilient AI-powered cyber defence systems, which can self-diagnose, defend and heal when encountering malicious or accidental attacks, or hardware and software anomalies.
In the fintech arena, AI designed by the D2 Lab has been able to learn customer behaviours and provide personalised recommendations for a micro investment application.
“In this case, two parallel neural networks were employed to learn about customer information and product item data for a final recommendation,” Yao said.
“Our industry partner has had considerable gains in both user positive feedback and profit.”
Yao and her team are also researching novel machine learning techniques that can take advantage of Internet of Things devices to accurately monitor and analyse human activities, in real-time, to support the aged care sector.
Creating a future world
It was Bill Gates’s book The Road Ahead that inspired Yao to devote her studies to computer science and engineering more than 20 years ago.
“I was fascinated by the future world he portrayed,” she said.
She has since won multiple awards for her work, including the 2020 UNSW Faculty of Engineering Excellence Award for Academic Excellence in the Education Excellence category.
Before joining UNSW in 2016, Yao was a lecturer and research associate at the University of Adelaide, where she gained both PhD and master’s degrees in computer science.
She previously worked as an engineer at telecommunications company China Unicom in Shandong after completing a master’s degree in electronic and communication engineering at Nanjing University of Posts and Telecommunications.