How do you start your day?
I have two daughters, aged 2 and 3, so as you can imagine my morning schedule is very demanding. It can start any time from 5:00am. After the situation is under control I get myself ready and go for a walk with my dog. These 20 minutes with the dog are very important for me to transition between the demands at home and my work.
Tell us about your roles?
My role at the University is split between teaching, research and administration. Teaching includes courses but also supervision of undergraduate and
postgraduate theses. Research is strongly collaborative. I mostly work with my PhD students but also with other international collaborators. In addition, much of my time is spent writing research grant proposals. Finally, as an academic I regularly participate in workshops and conferences to get feedback, as well as strengthen and broaden my network.
My role at the company is more practical. I am involved in investigating how AI components can improve the efficiency or add value to the process and products of the company.
How did you end up becoming an AI scientist?
I always found mathematics and computer science fascinating. Aged 18, I worked as a software developer in a bank where I mostly learned on the job. In the seven years I spent in that role I gained a lot of experience in programming with different tools and in parallel to that I read my undergraduate degree in
computer science. This is where I was introduced to neural networks. Neural networks are inspired by the processing in the brain and are capable of learning non-trivial relationships between many variables. For example, a neural network can ‘read’ the pixels of a face image and determine whether the face belongs to a man or a woman. In recent years AI scientists have constructed neural network architectures of many layers through sophisticated learning algorithms. This type of algorithm is known as deep learning. The idea of having algorithms that can learn in this way has always intrigued me. It is this passion that has brought me to where I am today.
What are some interesting projects you have worked on?
During my PhD, I worked on the development of novel trainable pattern recognition algorithms which could be used in various computer vision and signal processing applications (e.g. object recognition such as traffic signs in road scenes, image classification such as gender recognition from face images, and detection of repeated patterns in audio signals). That work was extended by four other PhD students with my collaboration.
At the moment I am working on the following projects*:
● Smart farming: In collaboration with two colleagues in Groningen and three partners at the University of Wageningen along with various stakeholders in the cattle and pig breeding industry, we designed predictive modelling techniques which would help farmers distribute the pigs in pens based on their expected growth rate, and help meat stakeholders to quantify the financial value of each pig. The algorithms use phenotypic data e.g. height and weight as well as depth images from a Kinect camera installed in the ceiling of an indoor pigsty.
● Forensic image analysis: In 2019, we were granted funding by the EU for the 4NSEEK project that deals with forensics against child sexual exploitation. The aim is to implement AI tools to assist law enforcement agencies in their investigations. Together with a PhD student and a collaborator from the University of Leon, I am responsible for determining whether two or more given images/videos were taken by the same camera device or not. We are developing a method that exploits the sensor pattern noise that is generated due to imperfections during the manufacturing process. Such noise is unique to every device, and its manifestation in the images/videos is not visible to the naked eye. I cannot overstate how valuable this is. The finding of a match will enable the police to drastically and quickly narrow down their search for offenders.
● Biomedical image analysis: I received internal funding from the University of Groningen on an interdisciplinary project in collaboration with the anatomy department of the medical sciences faculty. The project just started and its goal is to automatically identify tissue structures in very large images generated by electron microscopes of nanometer samples from the pancreas. This would allow medical experts to speed up diagnosis.
What do you see for the future of AI and should we be afraid of it?
In a way, this is similar to the first industrial revolution, where we moved from hand production methods to machines. It had improved the quality of life tremendously and opened up new disciplines and job opportunities. The applications of AI are unlimited. They can vary from an intelligent agent in
your smart home lighting system to computer aided diagnosis systems and the fully autonomous landing of a Boeing. Face detection is probably one of the
most popular AI-based applications, and you can find it in different products nowadays. With enough examples of face images, AI techniques can learn what are the features, their geometry and spatial arrangement that make up a face. Quite simply, there is no future without AI. Therefore, we should not fear it but embrace it.
You work in both academia and for a business, how different are these two professional environments and in what ways can they complement each other?
They are quite different but they certainly complement each other. Academic research is mostly curiosity-driven whilst businesses, especially the small to medium size, are product driven. I find this combination ideal, as I can keep up with both the real world needs and opportunities and the latest academic research, which can inspire new products.
Nowadays, it is increasingly important to have IT companies connected with the Universities. It is a win-win as it can be very beneficial when it comes to funding opportunities and sharing of resources. For instance, our masters program at the University of Groningen expects students to pursue an internship in the industry. Collaboration with companies plays a key role in this regard.
What are the challenges of working in a lab and managing a team?
At the University, I currently have a team of 5 PhD students with whom I conduct my research. Sharing individual progress within the team is probably what makes it challenging but fun at the same time. Before I embark on a big project with new collaborators, I start working on a small low-risk project. That experience then helps me to understand whether we have good compatibility. If so, we go for high-risk projects. Moreover, even if the short project goes smoothly, there is no guarantee that the success will be replicated in future projects. What is important is to keep effective and regular communication between all participants.
Did you ever face a dead end in your research, and how did you respond to it?
This can certainly happen. In that case I try to re-evaluate the research questions and the scope of the project, and discuss with my collaborators. While the pursuit of a research question may reach a dead end, the challenges you come across along the process usually serve to fine tune the research questions or
to shift the focus. Commitment, dedication, and enthusiasm are key factors to restoring faith in projects.
When was a moment you were most proud of and why?
When I became a father! It was a magical moment and the privilege that I feel as a parent makes me try to be a better person every day.
Name a book or podcast that has inspired you?
Rather than a book or podcast I would like to mention that I am very inspired by the life and writings of Daphne Caruana Galizia, an investigative journalist that was assassinated. Her courage and quest for truth are inspirational!
Which song makes you feel good?
There are quite a few, but “It’s a wonderful world” by Louis Armstrong probably tops them all.
What do you like to do for fun?
Playing football and flying my drone.
What inspiring quote would you like to share?
“I will not follow where the path may lead, but I will go where there is no path, and I will leave a trail”, Muriel Strode.
*The projects are in
collaboration with Prof. E. Alegre, Prof. B. Ducro, Prof. B. Giepmans, Prof. D.
Karastoyanova, Prof. N. Petkov, and Prof. R. Veerkamp, and the PhD candidates A. Alsahaf, A. Aswath,
and G. S. Bennabhaktula.