Combining mathematics and AI for smarter image processing

Can you tell us a little about your background? 

I have a background electrical engineering, telecommunications, information and communication technologies, and software and computer systems. I completed my bachelor’s degree in electrical engineering at Budapest University of Technology and Economics in Hungary, followed by a double master’s degree in Telecommunications Engineering and ICT for Internet and Multimedia at the Polytechnic University of Madrid in Spain and the University of Padua in Italy. 

Alongside my studies, I gained practical experience across telecommunications, data engineering, software development, machine learning, and IoT applications, which gave me a broad technical foundation across both software and communication systems. 

What led you to pursue a PhD?

I have always enjoyed understanding how things work at a deeper level and solving problems where there is not necessarily an obvious answer. During my master’s studies and previous technical projects, I became increasingly interested in research, particularly in the intersection of signal processing, optimization and machine learning. 

What motivated me further was seeing both the strengths and limitations of modern deep learning. Neural networks can achieve very impressive results, but they can also require large amounts of training data and computational resources, while often giving limited insight into how they arrive at a solution. 

I became interested in whether we can combine established mathematical and physical knowledge with learning-based methods to create models that are more efficient, interpretable and reliable. A PhD gives me the opportunity to investigate these questions in depth while contributing to research that can eventually have practical industrial applications. 

What attracted you to the IRS TransTech research school?

What attracted me most was the combination of strong academic research, international collaboration and industrial relevance. 

My project brings together several areas that I am very interested in, including signal processing, inverse problems, optimization, computational imaging and machine learning. At the same time, IRS TransTech provides an opportunity to work not only within Mid Sweden University, but also with international researchers and industrial partners. 

I particularly value this international aspect. I have studied and worked in several European countries, so I appreciate environments where people with different academic and professional backgrounds can exchange ideas and approach problems from different perspectives. 

For me, IRS TransTech offers the opportunity to develop as an independent researcher while building an international network and keeping the research closely connected to real technological challenges. 

Tell about your research interests and what you hope to explore during your PhD studies? 

My research focuses on Signals processing, computational imaging, inverse problems and model-based deep learning. I am particularly interested in how we can improve image reconstruction and processing in challenging conditions such as low light, blur, noise and limited resolution. 

During my PhD, I will be investigating how mathematical optimization algorithms can be transformed into dedicated neural network architectures through algorithm unrolling. The idea is to combine knowledge about the imaging system with machine learning, rather than relying on a neural network to learn everything from data alone. 

My goal is to develop models that are more efficient, generalizable and easier to interpret, while requiring less training data and computational resources. This is especially important for industrial imaging applications where processing needs to be fast and suitable for edge and IoT devices. 

Who are you going to collaborate with within the research school?

As part of my double-degree PhD, I will collaborate closely with Tampere University in Finland. This will give me the opportunity to work across two academic environments, learn from different research groups, and benefit from complementary perspectives throughout my PhD. 

I will also collaborate with Observit AB, the industrial partner in the project, which brings expertise in video surveillance, AI and edge computing. Together with the STC Research Centre at Mid Sweden University, this combination allows me to connect academic research with real-world industrial challenges. 

What are your first impressions of Sundsvall and Mid Sweden University, and what are you most looking forward to during your PhD?

My first impressions of Sundsvall and Mid Sweden University have been very positive. I enjoy the calm environment and the nature surrounding the city, and I have found Sundsvall to be a pleasant place to settle into. At the university, I have found the research environment welcoming, collaborative and international. 

During my PhD, I am looking forward to developing as an independent researcher and deepening my knowledge in computational imaging, signal processing and machine learning. I am also excited to learn from the researchers, fellow PhD students and industrial partners within IRS TransTech, and to benefit from the different perspectives that come with an international research school. 

I am particularly looking forward to the international collaborations and to seeing how the ideas developed during my research can move from mathematical models and experiments toward practical applications. On a personal level, I am also looking forward to experiencing more of Swedish culture, exploring Sundsvall and its surroundings, and making the most of my time in Sweden.

Read more about IRS TransTech