On 1 March 2023, Decebal Constantin Mocanu joined the Department of Computer Science at the University of Luxembourg as Associate Professor in Machine Learning.
Decebal Constantin Mocanu shares his background, former experience and explains his future challenges.
1) Could you introduce yourself?
“I am doing fundamental research in Machine Learning, particularly in Deep Learning, in synergy with all learning paradigms while also considering adjacent fields such as Evolutionary Computing or Network Science. Following a multidisciplinary approach, my team and I are working towards conceiving a new generation of versatile and adaptable Artificial Intelligence models while also taking into consideration their resource requirements and environmental impact for the benefit of science and society. I hold a PhD from TU Eindhoven, a MSc in Artificial Intelligence from Maastricht University, and a BEng in Computer Science from the University Politehnica of Bucharest. Previously, I have worked as an Assistant Professor in Machine Learning at TU Eindhoven and the University of Twente and as a software developer in industry.”
2) Why did you join the University of Luxembourg?
“The University of Luxembourg is a very young and dynamic university with high ambitions. It strives for excellence and aims to become a first-class, internationally recognised academic institution being well-supported by the government and society. This is also the perfect multicultural environment for me to develop my future academic career. At the same time, I am thrilled by the challenge of bringing my own contributions to the University’s aims and growth. From a personal perspective, I believe that Luxembourg has friendly people whom I am curious to know more about and a beautiful nature that I am eager to explore.”
3) What will be your main activities and challenges?
“I will split my time among the typical professor activities: research, teaching, and management. Research-wise, I will focus on improving Deep Learning versatility and lifelong learning capabilities in changing environments while decreasing computational requirements and energy consumption. The main challenge is that state-of-the-art Deep Learning techniques are widely successful just for single-task learning. Even so, they generate a severe waste of resources due to their dense connectivity, limiting at the same time deep learning access for a majority of possible users.
Currently, my main research area focuses on Sparse Training for Artificial Neural Networks which is a key element in alleviating all these issues. Based on the findings from my team and independent researchers, I extrapolate that significant advances in this area can create the next generation of AI techniques and systems which can have a much higher generalisation power than the current state-of-the-art. Moreover, they would be capable of learning a large number of tasks continuously in changing environments while being with orders of magnitude more efficient than the current state-of-the-art techniques. The latter would enable deep learning access to most users who don’t have large financial resources and can contribute to Greener AI techniques.
I enjoy teaching and working with students. From this perspective, I will focus on advising students to bring them to their best version of themselves by paying attention at the individual level. Also, I will design and teach new Machine Learning courses for the bachelor and master programmes. The main challenge is understanding the existing curriculum well to properly embed the new materials into it. From a management perspective, I will contribute to the department and university’s administrative needs. Nevertheless, I will also focus on building a strong and united research group in Machine Learning. We need to find good ways of collaborating and socially interacting together in a hybrid manner as the core group members are currently split among three universities (the University of Luxembourg, TU Eindhoven, and the University of Twente). Some affiliated members are even in distant time zones. All in one, there are many challenges to address and aims to achieve. Yet, I am looking with confidence to the future while being happy that I am here at the University of Luxembourg and eager to start working.”