42KL students do Malaysia proud, emerging runner-up in global ecological hackathon
By Digital News Asia July 13, 2021
- No prior knowhow of machine learning but 42KL’s learning model prepared them
- Only team with more accurate machine learning plankton identification model
[Article updated with quote from Jeff Sandhu, Head of 42 Kuala Lumpur & Sunway FutureX Talent.]
On the weekend of 26th & 27th June 2021, students, and data science enthusiasts from all around the world took on the first ecological hackathon organised by Veolia and the Tara Ocean Foundation – with the support of the oceanography laboratory of Villefranche, Sorbonne University and Plankton Planet.
The challenge? Participants had 48 hours to come up with a machine learning model that can classify hundreds and thousands of plankton photos taken during the Tara Ocean Foundation's current expedition (pic, below) around the world. Plankton are the principal producers of atmospheric oxygen and the vital first link in the oceanic food chain, thus are an essential component of life on Earth.
The hackathon placed a huge emphasis on energy frugality, whereby the model proposed must be able to run under low power utilising Raspberry Pi as the hardware component. Participants must also imagine the best way to visualise the data. Eventually, the model will be used on the Tara Schooner to work with the Planktoscope, the system developed by Plankton Planet and Stanford University to capture images of plankton in water samples.
Piscine bootcamp at 42KL changed way team members learnt, proved invaluable in competing in Veolia and the Tara Ocean Foundation hackathon
A team from 42 Kuala Lumpur (42KL), a computer-science programming school based in Sunway City, Selangor, Malaysia took on the challenge and went through a gruelling 48-hours, which were mostly spent on learning the fundamentals of machine learning. (pic collage from top right) Suresh Chetty, Hans Hazairi Baddrul Hezri, Gates Tan and Bunyod Shamsiddinov came from various backgrounds, and applied the same principles of 42KL's pedagogy during the hackathon.
Together, not only did they manage to secure a seat in the Top 8 finals, but they were also the only team that successfully developed a machine learning model with a plankton identification accuracy that is higher than that of the current model used onboard Tara Schooner.
"The challenge was immense as none of us had any prior exposure to machine learning, and I had never explored Python programming language. So, we initially just took it as a way for us to speed-learn machine learning models" said Gates who was a medical science graduate prior to joining 42KL.
By applying the same peer-to-peer learning pedagogy at 42KL, the team decided to learn as much as they could about machine learning prior to the start of the hackathon. "Right after we registered for the event, we studied various machine learning videos and tutorials, and we were confident entering the hackathon. But when it started, we quickly realised that everything that we had studied before was useless and we had to develop something much bigger than just a beginner model!" said Bunyod who, at 19 years old, is one of the youngest cadets (students) in 42KL.
"And you also have to be absolutely sure that the model you have developed will work because it takes time to train your machine learning model. So, there was a lot of going back to the drawing board for us," he added.
Fortunately for them, solving challenges with zero prior knowledge to begin with is what 42KL has prepared them for, right from the beginning. This, combined with their own unique skill sets, experience of working together, saw them discover an approach that could train their model to work with a variety of datasets.
According to Jeff Sandhu, Head of 42 Kuala Lumpur & Sunway FutureX Talent, "Our entire team here at Sunway iLabs are surprised by this achievement. What made this more special is that the team was made up of individuals who took part in our first ever Piscine bootcamp back in February 2021. Prior to this, they were just strangers among each other and that's what we at 42KL thrive upon. Bringing these likeminded individuals together from all walks of life and realising their potential by working collaboratively. We have high expectations when this first cohort is ready for their Industry Placement with our partners."
"The Piscine bootcamp at 42KL really changed the way I learn. It is a lot quicker for me to learn new stuff now and I just like how everyone at 42KL has the same spirit of learning," said Hans Hazairi, a former civil engineer who resigned from his previous role to upskill himself.
This sentiment was also echoed by Suresh, a former freelance Python & Django developer. "It's more collaborative. You bring what you know to the party, nobody needs to stand out and we collaboratively see how everything falls into place."
After completing the hackathon in 48 hours, the team was then invited to present their model in front of a panel of judges composed of Veolia and Tara employees. While the team had the most accurate machine learning model, they lost the top spot due to the energy consumption needed to operate their model in the middle of the ocean with no internet access.
"They were very transparent with us on where we did not do as well as the top placed team, but we were just surprised that we got this far," said the team in a joint statement.
The team is now currently focused on improving their machine learning model in hopes that it could be applied into other verticals such as agriculture where food security is a major concern here in Malaysia.
"In the future, we are hoping to apply this model and our learnings from it to be used for other types of identification such as the health of our plants and vegetables," said the team. One future project could be a collaboration with Sunway FutureX Farms, Malaysia's first innovative urban farm R&D hub, producing fresh, pesticide-free vegetables using sustainable and safe farming technology.
The official organiser press release can be read here.