This year's winners of the Student Scholarship Program include doctoral students from our Faculty. Dominika Kunc and Albert Sawczyn showed exceptional achievements in their fields. A total of eight WUST students received the scholarships during the Science Day in Wrocław. Congratulations!
The Student Scholarship Program is an initiative of the Mayor of Wrocław and is run by the Wrocław Academic Centre. Applications submitted by doctoral students from Wrocław universities (in eight categories) are evaluated by committees including recognized authorities associated with the Wrocław scientific community. The amount of the scholarship is PLN 2,000 per month, and it is paid to the awarded students for a maximum of nine months.
Dominika Kunc
The supervisors of her doctoral dissertation titled „Representation learning for physiological signals collected in real-life using wearables” are Prof. Przemysław Kazienko and Stanisław Saganowski, D.Sc. (both from the Faculty of Information and Communication Technology).
–In psychology, there are various models of basic emotions, such as Ekman and Friesen's six emotions or Plutchik's eight opposing emotions,” says Dominika Kunc. However, none of these models provides a definite answer to the question of how many emotions people actually experience, or how different affective states are reflected in human physiology".
That is why, as part of the LarField study conducted with the Emognition team, Dominika Kunc collected unique data on emotions in everyday life, using modern wearable devices. – The study was performed on 167 residents of Wrocław, who collected physiological data (electrocardiography (ECG), blood volume (BVP)) for a month and regularly filled in affective questionnaires – says the scholarship holder.
The result is one of the world's largest data sets on emotions in everyday life, covering a total of about 4 years of physiological recordings and over 18,000 samples with affective annotations.
– Physiological signals provide valuable information about our emotional states, but their traditional analysis is time-consuming and complicated – explains Dominika Kunc. Therefore, it is of key importance for the project to use advanced representation learning models that transform raw data into a compact, informative form. These representations allow the identification and analysis of patterns that may be difficult to detect using traditional methods.
– My research focuses on the exploratory analysis of clusters of psychophysiological states using unsupervised machine learning and representation learning techniques – says the scholarship holder. – The goal is not only to discover new patterns, but also to examine their variability within and between individuals, which could allow a deeper understanding of emotions in their natural context.
By using dimensionality reduction techniques, it is also possible to visualize these representations and create an easily interpretable psychophysiological map of affect.” Combining innovative machine learning techniques with the unique LarField dataset not only expands our knowledge of human emotions, but also opens new possibilities for practical applications, such as monitoring mental health and supporting mental well-being in society.
– The application could be for example used to monitor the affective states of people suffering from depression and predicting when a depressive episode will occur. As a result, it will be possible to take action well in advance – explains Dominika Kunc, who currently also holds a scholarship under the Opus grant „Affective informatics: deep learning in recognizing emotions from physiological signals".
She also completed research internships at the University of Southern California (USA) and at Nanyang Technological University (Singapore).
Albert Sawczyn
Albert Sawczyn is preparing his doctoral dissertation „Methods of reasoning on knowledge graphs" under the supervision of Prof. Tomasz Kajdanowicz at the Faculty of Information and Communication Technology.
– In my research, I focus on reasoning methods, and in particular on those that use knowledge graphs as a source of information – explains Albert Sawczyn.
In the face of the growing demand for intelligent, knowledge-based systems, such as chatbots, knowledge graphs have particular advantages. By showing information in the form of relationships between objects, knowledge graphs offer a more condensed and transparent way of presenting knowledge. – Unlike text documents, these graphs can be easily updated and modified – explains the student from WUST. – This is particularly important in dynamically changing fields, such as medicine and politics.
The research achievements of Albert Sawczyn are so far related to learning graph representations and natural language processing. “I am currently working on developing methods for detecting hallucinations generated by large language models (LLM) in cooperation with the University of Technology Sydney, adds the scholarship holder.
Gallery
