Professor Dacheng Tao, Director of the UBTECH Sydney Artificial Intelligence Centre has focused his research on the intersection between artificial intelligence and data mining, with his most recently awarded work having achieved an effective, robust and succinct representation of original high-dimensional, noisy and unstructured raw data.
Representation learning is essential because real-world data such as images, video, and text are mathematically and computationally inconvenient for specific machine learning tasks, such as classification and clustering.
“It’s an honour to be awarded an IEEE ICDM research contributions award," said Professor Tao, from the School of Computer Science.
"Not only has it recognised my effort and contributions to date, it has also strengthened my commitment to research in data mining and artificial intelligence, in particular representation learning.
“The field of artificial intelligence is blossoming; it’s driving both economic growth and social progress in what is now the world’s fourth industrial revolution.”
Having recently completed an ARC Future Fellowship, Professor Tao’s recent research has concentrated on developing multi-view learning, a process which allows machines or robots to consider data from multiple data sets, forming a more comprehensive view and allowing them to behave accordingly.
“Multi-view learning is essential for intelligent systems, in that those systems need to complete tasks based on the input data from different sensors," explains Professor Tao.
“We have developed algorithms that can learn an effective integrated feature from different sources for the subsequent processing, such as classification, with theoretical guarantees.
“I’m now leading a team of early career researchers and research students to work on my ARC Laureate Fellowship. We aim to devise a suite of algorithms for processing and understanding videos captured by moving cameras, and to establish mathematical foundations for deep learning-based computer vision.
“The project has already yielded considerable results — such as explaining why deep learning is superior to shallow learning and recovering depth information from a single RGB image."
Rather than developing artificial intelligence to replace humans, Professor Tao instead believes its primary purpose is to augment the human experience, filling in gaps or performing tasks that humans are incapable of doing.
“My team is working to ensure our developments will enhance humankind’s capability, increase humans’ productivity and improving our quality of life,” says Professor Tao.
“Already we have developed algorithms to track human pose from videos and thus to help robots mimic human’s activities.”
Having authored over five-hundred papers on such topics as machine intelligence, pattern analysis, and image processing, Professor Tao’s work has also been cited over 36,000 times within both the fields of Engineering and Computer Science.
Working with his fellow researchers and students, Professor Tao is adamant his recent achievement is the result of a group effort.
“I would like to extend sincere appreciations to my students, fellow researchers, collaborators and colleagues who have worked with me over the past years – this recent award just as much is about their work as it is mine,” says Professor Tao.