The disruptive nature of smart technology

A researcher of international standing, Dr Alexandre Mendes is using machine learning to solve real-world problems, helping to revolutionise decisions, processes and even diagnoses.

Image of Alexandre Mendes

Artificial intelligence (AI) is often touted as the future of technology. But what exactly is AI and why is it so exciting for businesses, industries and governments alike?

Dr Alexandre Mendes’ research focuses on the AI area of machine learning, which is when computer systems identify data patterns, learn from data and solve problems with minimal human involvement.

Machine learning is driving new technological advancements every day across almost every industry. It is helping diseases be identified earlier, driving new ecommerce solutions and facilitating novel voice search applications. Over his career, Alexandre has utilised machine learning for a range of practical outcomes. Most recently, his work is exploring how deep learning and optimisation—two areas of machine learning—can improve the efficiency of Australia’s transport and energy industries.

“My work focuses on artificial intelligence, but that is a huge research area. More specifically, I work in the two sub-areas of optimisation and deep learning.

“Optimisation involves the use of software to solve complex problems that require decision making. Deep learning involves the use of deep neural networks, which mimic the structure of the brain, for complex classification and prediction problems.

“The use of these complex computational tools impacts so many different areas. We have so much local knowledge, so much potential, and so much room for improvement that the opportunities are endless.”

On track to success

Central to machine learning is optimisation. This complex process is essentially about finding the best solutions to a problem by analysing data, switching tactics when needed and minimising errors. In partnership with the Hunter Region rail network, Alexandre is researching how this clever technique can be applied to coal train logistics.

“I am currently working with train logistics, helping to push as many trains through a rail network as physically possible, without violating operational and safety constraints.

“That is a very difficult problem because most of the network consists of single tracks. Therefore, trains travelling in opposite directions must negotiate who will stop at the few available side tracks so the other trains can pass through. When you are trying to push as many trains as possible through the system, this coordination becomes critical.”

Alexandre’s high-level goal is to create a method that can schedule train trips in real time through the rail network. This will help to get the best performance out of the existing infrastructure—boosting efficiency while keeping costs at a minimum.

“Better train coordination affects everyone. The same knowledge used for coal train coordination in the Hunter Region can be used for better coordination in, say, the Sydney Metro, or the intercity trains. But more short term, having a more efficient supply chain for coal export in Newcastle generates more revenue and jobs.”

Powering the future

Another aspect of Alexandre’s work involves researching the practical applications of deep learning. Part of machine learning, this process allows computers to self-learn using data examples.

Teaming up with CSIRO’s Newcastle energy centre, Alexandre is using this deep learning process to predict solar energy output. The AI technology analyses extensive reams of cloud cover images and weather information, such as wind speed and direction, then, based on this data, learns how to predict short-term solar generation.

“Humans are very good at guessing the configuration of the clouds in the sky in the future, after just a few seconds observing how they move. But for computers, that is still a very difficult task.”

Alexandre explains that determining how much sunlight will reach a solar farm within a few minutes is crucial for better coordination between highly variable renewable generation and more stable base generation based on coal and gas.

“Solar generation prediction will help maintain power supply quality and better integrate renewables into the generation mix. The high-level goal is to predict how much energy a solar farm will produce in, say, the next 20 minutes, so that the transmission/distribution network can adjust beforehand for the variation in generation.

“This research will affect anyone with a solar panel on their roof who wants to integrate their excess generation with the grid, and the companies that run the system.”

Smart, industry-focused solutions

Alexandre’s research is always conducted in collaboration with industry partners, allowing results to translate into immediate solutions.

“By working directly with companies, any positive result from our research can be applied immediately and influence how business is conducted in organisations.

“It can be challenging sometimes to generate enough interest and confidence from industry, so that companies want to invest in this kind of applied research. But once they do, they realise the immense potential that the research has to offer their business.”

The highly applicable nature of machine learning across industries has put Alexandre and his team in hot demand. They are always keen to take on new PhD students to help with the work, and find inspiration in helping the next generation test new disruptive technologies that could facilitate better processes and outcomes for people.

“The sheer amount of research that is generated by the School of Electrical Engineering and Computing means we are always looking for PhD students with a computer science or software engineering background. Having students conducting research in such areas also means that they can see the results of their work being used immediately.

“Personally, it is great to see students grow throughout the process and finish their PhD as motivated, competent researchers that, in many cases, end up being employed by the partner company to continue the research work internally.

“At the University of Newcastle, we are committed to producing the highly skilled workforce that the world needs so much in this era of constant innovation and incredibly fast technology change.”

The University of Newcastle acknowledges the traditional custodians of the lands within our footprint areas: Awabakal, Darkinjung, Biripai, Worimi, Wonnarua, and Eora Nations. We also pay respect to the wisdom of our Elders past and present.