You are here

Emotion Detection and Artificial Intelligence

22 February 2016
Photo of Prof Roland Goecke, University of Canberra

Having flexible high performance computing (HPC) architecture that is connected and collocated with data storage is a critical service combination that enables the research into the detection of depression by researchers in the Human-Centred Technology Research Centre, at the University of Canberra.  The psychologists, linguists, performance analysts, and computer engineers, in the Research Centre rely upon the eResearch services provided by Intersect to store their data and access HPC services at the National Computational Infrastructure (NCI).  The audio-video data of 150+ study subjects (roughly 50 terabytes in size) is cross-linked with electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI) data and is analysed simultaneously.  

Intersect's range of services is increasingly important as the reliable means for researchers to focus on their advancing their research in shorter timeframes and enables the next generation of researchers (higher degree research students) in the Research Centre, to learn new data and technology intensive skills.  By example, in an international breakthrough, Professor Roland Goecke, Professor of Affective Computing at the University of Canberra and his team of researchers have shown that their new technique is able to correctly detect depression with an accuracy of up to 95%. This approach could expand into other disorders (schizophrenia, autism, bipolar disorder), where laboratory-style diagnoses are also lacking.  

As the project scales up with data from 150+ to 1000 study subjects over the next 2-3 years, so will the research infrastructure, enabling vital cross-modal and longitudinal temporal information to be studied.  

Read a news report about this research that has been undertaken at University of Canberra, in the Canberra Times.