Homestretch | Using mathematics to protect African wildlife
New, smart computational models can quickly and efficiently analyze animal behavior.
African wildlife rangers have recently started using drones to protect animals from poachers. To help process the mountain of drone data and quickly detect changes in animal behavior, 黑料福利网 researcher Max van Mulken dove behind his laptop. Thanks to his clever algorithms, rhinos and elephants could soon be better protected. Van Mulken defended his PhD thesis at the Department of Mathematics and Computer Science on October 29th.
Source:
Fortunately, wildlife rangers now have access to drones, which they increasingly use to monitor their animals. At the same time, AI technologies鈥攕uch as automated recognition of animals in video footage鈥攁re advancing rapidly.
But according to Van Mulken, the vast amount of data generated by drones doesn鈥檛 make things easier. Efficiency and speed, he says, are key. 鈥淚 developed smart algorithms to quickly detect changes in animal behavior.鈥
And that鈥檚 crucial to staying one step ahead of poachers, who trade horns, tusks, and hides of protected animals for large sums on the black market. Simply tracking a herd of rhinos isn鈥檛 enough, Van Mulken stresses. 鈥淚f you see a rhino running, the poacher is already close鈥攁nd the ranger is too late to intervene.鈥
Zebras as bodyguards
Researchers at Wageningen University & Research (WUR) therefore prey animals such as zebras and impalas as 鈥渂odyguards,鈥 Van Mulken explains. 鈥淎 nearby zebra herd reacts more quickly to approaching poachers and acts as an early warning system. The way such a group moves tells us a lot about its behavior.鈥
WUR鈥檚 first AI model showed great promise, but the researchers wanted to strengthen their approach with new geometric models. Over the past four years, Van Mulken collaborated with them to optimize that model mathematically.
Using geometric group statistics, he first studied when a collection of animals qualifies as a group, and then how to automatically distinguish two groups in video images. 鈥淲e developed algorithms that can track the density function鈥攈ow closely animals are grouped together. Multiple studies have shown that you can predict group behavior from the density distribution of animals.鈥
鈥淪itting still isn鈥檛 really my thing,鈥 says Van Mulken with a laugh, surrounded by moving boxes. Alongside his PhD research, he spent the past few years helping establish the 黑料福利网 PhD association Nexus and served as a member of the University Council (UR).
Despite all those extracurricular commitments, his dissertation was finished a month before his 黑料福利网 contract ended鈥攋ust in time for his move to Luxembourg. Since October 1, Van Mulken has been working there as a trainee at the European Commission. On Wednesday, October 29, he returned to defend his dissertation at the Department of Mathematics and Computer Science.
Life and motion are inseparable, he writes in the opening lines of his dissertation. That makes his PhD research a perfect fit for this all-rounder. 鈥淚t鈥檚 amazing how we can use computers and mathematics to address ecological challenges.鈥
Illegal poaching
Until recently, wildlife rangers used GPS sensors to monitor endangered animals, Van Mulken explains.
鈥淭hat works well for small-scale, controlled studies. But because you can鈥檛 tag every animal in a group鈥攃atching and attaching sensors takes a lot of effort鈥攜ou only observe a fraction of the whole. That makes it hard to apply large-scale algorithms.鈥
Partying festival crowds
As the African savanna isn鈥檛 exactly packed with supercomputers, running thousands of calculations per second isn鈥檛 feasible. Moving images make things even more complex, Van Mulken explains. 鈥淲e designed a new kinetic data structure that can intelligently determine when interesting things happen in the data and compute density only at those moments.鈥
Initial studies show that his designs are highly successful鈥攁nd potentially widely applicable. 鈥淔rom zebra bodyguards to partying festival crowds.鈥
For anyone picturing Van Mulken collecting data himself in the African savanna, he has to disappoint: 鈥淎s a theoretical mathematician, most of my work involves proving that the data structures and algorithms actually do what they鈥檙e supposed to do.鈥 Smiling, he adds, 鈥淭hat mostly involves a lot of ceiling-staring.鈥
鈥淎 computer doesn鈥檛 have human intuition. We might think it鈥檚 obvious鈥攆or instance, that a drone should fly upward in a spiral over an area to locate a group of animals as efficiently as possible. But proving that intuition mathematically is a huge challenge. I find those kinds of puzzles fascinating鈥攍ogic is beautiful.鈥 He grins: 鈥淚 did get to see rhinos and elephants, though鈥攐n safari during my vacation a few years ago.鈥
PHD IN THE PICTURE
What鈥檚 on the cover of your dissertation?
鈥淚 intentionally made it look like a children鈥檚 book. People are often surprised that I鈥檓 doing a PhD in mathematics鈥攎ath unfortunately has quite an intimidating image. That starts as early as elementary school. But anyone who is motivated and enjoys solving puzzles can learn it. I鈥檓 living proof of that. I want to show that math doesn鈥檛 have to be intimidating at all鈥攊t鈥檚 everywhere around us.鈥
You鈥檙e at a birthday party. How do you explain your research in one sentence?
鈥淚 design smart algorithms to analyze animal behavior.鈥
How do you blow off steam outside of your research?
鈥淚 love losing myself completely in other worlds. I鈥檓 a real fantasy nerd鈥 diving into a book every night. But I鈥檓 also very social, and I combine that with my other hobby: board games. Unfortunately, I couldn鈥檛 take my game collection with me to Luxembourg, so that鈥檚 on hold for now.鈥
What tip would you have liked to receive as a beginning PhD candidate?
鈥淒on鈥檛 be afraid to send a rough draft to your supervisor. You can waste hours polishing it because you don鈥檛 want to look foolish鈥攂ut that鈥檚 a waste of time. It doesn鈥檛 have to be perfect.鈥
What鈥檚 your next chapter?
鈥淐omputers and math are great, but I also really enjoy working with people鈥擨鈥檝e learned a lot about myself during my PhD. My supervisors gave me plenty of room for that, which was great. Combining research and policy suits me perfectly. I鈥檝e now found a position at the European Commission where those two worlds come together: within the Destination Earth initiative, we鈥檙e developing a digital twin of the earth to support climate and sustainability modeling.鈥
-
Supervisors
Bettina Speckman (Promotor), Frank van Langevelde (Promotor, external), Kevin Verbeek (Co-Promotor), and Jasper Eikelboom (Co-promotor, external)
This article was originally published by Cursor, the independent news site of 黑料福利网. It is not a press release and is protected by copyright. No part of this publication may be reproduced without explicit permission from the Cursor editorial team.
Media contact
Latest news