Shanshan Wang’s doctoral dissertation is both timely and relevant because it focuses on self-supervised learning. This technique allows systems to learn patterns and make decisions by analyzing natural relationships in data.
“Just like children learn by observing and listening, intelligent systems can learn by finding alignment between what they see and hear,” Wang explains.
Her work is especially valuable in fields where labeled data is scarce or expensive to produce, such as environmental monitoring, public safety, and autonomous driving. Wang’s research introduces a new urban audio-video dataset, improved training techniques, and novel methods that help machines focus on meaningful sounds and visuals while ignoring noise. These techniques significantly enhance a system’s ability to recognize and classify scenes, even with minimal training data.
This research connects to ongoing public discussions about ethical AI, data efficiency, and responsible automation. As society increasingly relies on AI in everyday life, from smart devices to autonomous vehicles, learning without needing human labels can help build systems that are more efficient, scalable, and broadly applicable—while respecting privacy and reducing the demand for human labor in data annotation.
Shanshan Wang conducted her doctoral research in Tampere University’s Machine Listening group and is currently working at CSC - IT Center for Science.
Public defence on Monday 09 June
The doctoral dissertation of M.Sc. (Tech.) Shanshan Wang in the field of Computing and Electrical Engineering titled Self-supervised Representation Learning on Audio-Video Data will be publicly examined at the Faculty of Information Technology and Communication Sciences at Tampere University at 12:00 on Monday 09 June 2025. The venue is in the auditorium Pieni Sali 1 of the Festia building on the Hervanta campus (address: Korkeakoulunkatu 8, Tampere). The opponents will be Senior University Lecturer Jorma Laaksonen from Aalto University and Research Scientist Xavier Alameda-Pineda from Université Grenoble Alpes, France. The Custos will be Associate Professor Annamaria Mesaros from the Faculty of Information Technology and Communication Sciences at Tampere University.
The doctoral dissertation is available online.
The public defence can be followed via remote connection.