“The collaboration with research engineers and data experts has made this a much more interesting research project”
What does research data management mean in everyday practice? In this series of interviews by RDM Support, researchers share their experiences with various aspects of research data management. In this interview, ecologist Joeri Zwerts tells about how he uses machine learning to take his research to a higher level.
Joeri Zwerts explains: "Sustainable, or responsible, forest industry is often recognized by the Forest Stewardship Council (FSC) label. But does FSC certification actually have a positive effect on the fauna in the forests concerned? Although very little is known about this topic, my research aims to answer this question". Joeri Zwerts is a PhD student at the Department of Ecology and Biodiversity in the Faculty of Science. His research data consists of video recordings of camera traps and sound recordings from the jungle of Congo, Cameroon and Gabon. To analyze the jungle sound recordings he has enlisted the help of, among others, the Research Engineering team of RDM Support. "You can get a lot of information from those sound recordings. But it is not possible to listen back and manually peek when you hear a monkey or other animal. The Research Engineering team helps me set-up a machine learning application that does this work for me."
By working together we have been able to realize much more than we could have ever done separately. That's why I encourage interdisciplinary work enormously
Monkey sounds and machine learning
“Sound recordings have a much wider coverage than a camera trap. When an animal is sitting in a tree you often don't see it, but you can hear it. Research into biodiversity using sound recordings (acoustic monitoring) is still an underdeveloped technique, especially in places where there is a lot of ambient noise such as rainforests. In order to carry out this research, you need an interdisciplinary approach. I am an ecologist, not an expert in machine learning. You need research engineers and data science experts who know about these techniques. By working together with RDM Support and the focus area Applied Data Science we have been able to realize more than we could have ever done separately. That's why I encourage interdisciplinary work enormously. I now work with machine learning models and the research engineers and data experts work with monkey sounds. These are two worlds that don't meet very often, but when they do, they reinforce each other; and we end up learning an awful lot from each other."
“A self-learning algorithm is required for this application. You need to train this algorithm, and for that you need training data. For this purpose, I made sound recordings in sanctuaries, and installed sound recorders in the cages to collect many vocalisations per species. Of course it would be great if this method could be used by more people, and not just me, because the possibilities are endless. Not only for researchers who want to investigate biodiversity, but also for those involved in eco-surveillance who wish to recognize the sound of a poacher’s gunshot for example. That is why we want to publish and make this method open source, so that other parties can plug in their own training data and use this application for another purpose.”
Support pushes research to the next level
“It would be ideal if every research group had someone available full-time to supervise code writing and to look at and think about the possibilities of research projects. This has the potential to increase the amount of information you can obtain from your research data. As a starting researcher you do have some knowledge of statistics and programming, but you're not an expert. During your career you might run into a problem or question that is complex and therefore your limited knowledge, or the knowledge of your research group, may not be sufficient. This certainly applies to rapidly developing techniques such as machine learning. However if someone with research engineering expertise is present, you as a researcher, no longer have to be limited by insufficient skills or knowledge about these methods. RDM Support can provide an added value with this type of expertise.”
Applied Data Science
This research project is co-funded by the focus area Applied Data Science. Twice a year there is a call for the Applied Data Science grants. These are small research projects of up to 5000 euros that are aimed at helping researchers to apply data science techniques in their own research. There is interdisciplinary collaboration between faculties and with RDM Support. For more information about these calls and an overview of the current projects, check the website of Applied Data Science.
Research Data Management Support
Have you become interested in the services of the Research Engineering team? Or would you like to read another RDM story about how the Research Engineering team helped another researcher? Take a look at our website, or contact us.
Do you want to see Joeri in action in Congo? Take a look at NPOplus for Helden van de Wildernis (Dutch only).