Less than a year after Dr Jack Yang turned his machine learning knowledge to the health sector, he has taken out a top award.
His paper on breast cancer data was judged the best paper at the 16th Australasian Data Mining Conference in Bathurst.
The conference covers a wide range of topics in computer science and data mining and had 90 papers submitted.
Of these, just 23 were accepted for the conference, and Dr Yang’s paper was judged the best.
“The criteria for success are readability, novelty and making a significant contribution to the area,” Dr Yang said.
The break-through is to use machine learning technology to predict breast-cancer patients’ survival rate using imperfect data.
Dr Yang has developed a novel imputation algorithm to intelligently replace missing values.
Experimental results demonstrate a significant improvement on the survivability prediction, compared to existing methods.
The strength of the research is significant validation for SMART’s new research structure – one of which focuses on health and the benefits possible by bringing modelling technology to the sector.
“We would now like to write a journal paper out of this work,” Dr Yang said.
“Still in the health sector, we can bring this approach to a wide range of different diseases.
“The importance of this one is that, once we have a model, we are able to provide better professional advice to the medical profession.
“We are able to better rank and determine which are the important factors that contribute to the survival of patients.”
Dr Yang is responsible for translating conceptual models into implementation programs and code prototyping at the SMART Infrastructure Facility.
His recent research involves new machine learning algorithms for big data processing. His team has developed Apache Spark-based platforms for analysing user networks and online behaviour modelling. Dr Yang is also leading a Discovery Project from the Australian Research Council (ARC).