This DSTA scholar learnt cutting-edge tech in AI and computer vision to create algorithms that enhance national defence

Hostile enemy soldiers camouflaged amidst dark foliage. Darting vessels on the horizon used by terrorists, indistinguishable in the night. Weaponised drones approaching too fast for defensive fire.

These are just some scenarios where the human eye may find it difficult to detect and pinpoint targets. With Singapore’s security at stake, enter the latest military technology developed for the Singapore Armed Forces to acquire and neutralise targets before harm is done to soldiers and civilians.

“Deep learning is a method in artificial intelligence (AI) that trains computers using a data-driven approach, in a similar manner that emulates how the human brain would learn from its experiences and mistakes,” explains Mr Lua Ying Hao, who was a recipient of the Defence Science and Technology Agency (DSTA) Scholarship in 2016.

More on this topic Deploying effective comms out in the field: How her work gives the military a boost Using her problem-solving skills to create cutting-edge defence technology

Deep learning is used to support “automatic target detection in complex environments for enhanced situational awareness”. He adds that this saves time and resources.

Today, as a senior engineer at DSTA’s Digital Hub, Mr Lua is involved in the development of deep learning solutions for computer vision applications.

“It is my honour and privilege to be able to contribute as one of the first few engineers to deploy and operationalise deep learning solutions in support of Singapore’s national defence,” says the 28-year-old.

Enhancing the nation’s armour

Mr Lua developed an early interest in AI when he marvelled at optical character recognition – how computers can automatically “read” pictures as text. This made him curious about how the technology works and its application in computer vision.

Wanting to find out more, Mr Lua interned at DSTA’s Digital Hub in 2018 where he worked on AI technology. “The internship was truly a launchpad for me. It was the first time I fully grasped the concept and implemented deep learning algorithms to train AI modules for various recognition and detection tasks,” he recalls.

The DSTA Scholarship gave Mr Lua the opportunity to study abroad. He graduated in 2020 with a Master of Engineering in Electronic and Information Engineering from Imperial College London.

In London, a network of seniors from DSTA shared tips and advice on living and studying abroad.

“These fellow scholars helped me to adapt to university life quickly, allowing me to focus on my curriculum,” says Mr Lua, who consistently topped the dean’s list.

At work, his colleagues are always willing to lend a helping hand. “Everyone at DSTA is really open and approachable, and you can always count on the support of your colleagues to get the help you need,” he says.

“It is my honour and privilege to be able to contribute as one of the first few engineers to deploy and operationalise deep learning solutions in support of Singapore’s national defence.”

– Mr Lua Ying Hao, recipient of the DSTA Scholarship

With deep learning advancing at a breakneck pace, it has been a steep learning curve for him.

“Deep learning has only become relatively viable in the last seven years. Back when I was in school, AI courses were not widely available or even offered. I had to learn on the job or during my free time,” he says.

“At DSTA’s Digital Hub, for example, we have several AI collaborations with top-tier institutions such as the Massachusetts Institute of Technology, Carnegie Mellon University, National University of Singapore and Nanyang Technological University, to name a few,” he shares. “These collaborations allow us to constantly learn from the best and adapt solutions from the leading researchers in the field.”

AI and deep learning-based algorithms “have outperformed traditional computer vision on image classification benchmarks like ImageNet, and detection benchmarks like COCO”.

He adds that this translates to more reliable algorithms so that Singapore’s military operators can be relieved from more labour-intensive work and be reallocated to more cognitively intensive work that brings greater value, such as having more bandwidth to come up with better countermeasures to defuse certain threats during time-critical windows.

Navigating global challenges

Under the scholarship, there is also a Global Internship Programme which allows scholars to intern anywhere.

Mr Lua decided to return to Singapore during a summer break at Imperial and did a three-month internship at software analytics company SAS in 2019. There, he worked on machine learning solutions for satellite image segmentation and generation of image captions.

He headed back to London to complete his studies after his internship, but the Covid-19 pandemic forced him to return home in March 2020, just before he was due to graduate in June that year.

He completed his degree in Singapore through online classes, examinations and presentations.

Looking back at his scholarship experience, Mr Lua encourages students to persevere through challenges and keep going.

“Always have faith in yourself and never give up,” he says. “Always dare to try new things and don’t be afraid to make mistakes and learn from them.”

Back to main page