Defence signs Deakin, UniSA for AI projects

The Department of Defence has signed contracts with two universities to help develop its AI capabilities.

The University of South Australia and Deakin University will receive a combined $1.75 million from Defence’s next generation technologies fund (NGTF), to work on the Defence Artificial Intelligence Research Network (DAIRNet) phase two.

UniSA’s project will be to develop a statistical machine learning algorithm to detect infection in a person, using data from devices like smart watches.

DAIRNet director Mel McDowell told iTnews that the research team "will be inoculating people with the flu vaccine."

"That’s a known time-point that people will be receiving something that will make their body react," McDowell said.

“They collect data of the individual up to when they receive the vaccine, they give the vaccine, and then follow that individual afterwards.

"They want to use that data to see if they can predict certain outcomes.”

This could be expanded in the future to include early detection of chemical or biological threats, to improve the effectiveness of interventions.

Deakin’s applied AI institute will use machine learning to “develop models that can process noisy and dynamic data that is multi-source, multi-modal, irregularly timed and that spans a prolonged period.”

The broader project Deakin and UniSA are working on aims to use AI “to process noisy and dynamic data into information that will give military decision makers an edge,” Defence said in a statement.

McDowell told iTnews ML works best with clean and predictable data, so it can “readily develop the tools and learn ... that makes it a robust model.

“The Deakin team is working on a new type of ML, working on data that is not clean, real-world data”.

Associate professor Truyen Tran at Deakin University’s Applied Artificial Intelligence Institute said: “By working closely with Defence experts, Deakin University is aiming to create a versatile AI technology capable of automating multiple tasks that process complex, time-based data.

“This data is often noisy, collected from multiple sensors with varying characteristics that are often operating under different conditions.”

Chief defence scientist professor Tanya Monro said the two-year projects will develop “prototypes that deliver Defence capabilities.”

“Robotics, autonomous systems and artificial intelligence are a group of technologies that are a Defence sovereign industrial capability priority,” she said.

“The DAIRNet phase two callout sought innovative proposals for prototypes that will help warfighters achieve superior decision making, and ultimately enhance Defence capability.”

DAIRNet was launched in November 2021 with the aim of bringing together a community of AI researchers working on novel AU technologies for Defence.

Its initial members were Monash University, RMIT University, Swinburne University of Technology, The University of Adelaide, UNSW and University of South Australia.

Its role is to manage research calls with Defence, provide strategic advice about AI technology developments, and develop Defence AI researchers.

Prior to today’s announcement, DAIRNet had launched 26 projects since its inauguration.