From automated DJ mixing to artificial intelligence in news media
NEWS
This Monday, no less than three doctoral students presented their work at the Department of Computing Science. Guests and opponents congratulated Mickaël Zehren, who succesfully defended his work and now can call himself a doctor. Sourasekhar Banerjee, who nailed his thesis, and Igor Ryazanov, who is halfway through his doctoral studies.
Mickaël Zehren was brilliant during his defence and now holds the title of Doctor. ImageZehren
Uniting computer science and music
In his thesis, Mickaël Zehren studies ways to automatically create DJ mixes by identifying points of interest in music tracks, known as cue points, which are essential for DJ mixing. The four-year thesis is a true interdisciplinary project that successfully brings technology and the humanities together.
"Mickaël Zehren delivered a nice and multi-faceted presentation, discussing music and algorithms", Professor Paolo Bientinesi says.
"For the past 4 years, Mickael acted as the link and the interpreter between Marco and myself, as our expertise and language are entirely different."
Three-way collaboration
Mickaël Zehren also contributed with his knowledge of Machine Learning (ML), together with a real passion for curating and sharing data. "I can safely say that all three of us learned a lot from this three-way collaboration," Bientinesi says.
A new approach to train AI models
At the same time, Sourasekhar Banerjee nailed his thesis ‘Advancing Federated Learning: Algorithms and Uses-Cases’, in Umeå University Library. Present were principal supervisor Monowar Bhyan, Professor Erik Elmroth as well as colleagues from the department and the Autonomous Distributed, Systems Lab research group. Sourasekhar Banerjee studies how to train and use AI models without moving data from end devices. "His work will ensure privacy while providing cost-effective solutions to users," supervisor Monowar Bhyan says.
Find out more when Sourasekhar Banerjee defends his thesis later this month.
Doctoral student Sourasekhar Banerjee at his nailing ceremony.
ImageVictoria Skeidsvoll
Detecting power structures in text
This afternoon, guests were also introduced to the interesting work of Igor Ryazanov, who is halfway through his doctoral studies. Ryazanov's research involves using automatic tools to better understand how characteristics such as initiative and responsibility are expressed in the news media. A headline like, ‘Will AI take your job?' for example, suggests that AI has a greater capacity for initiative than the related headline; ‘This is how AI will affect the future labour markets'.
"The same methods can then be used to explain how other types of power structures are expressed in text," says associate arofessor Johanna Björklund, principal supervisor. She looks forward to following Igor Ryazanov's further work.
The Department of Computing Science at Umeå University has grown at a record speed and is today Sweden's second-largest department in this field. Researchers from all over the world are gathered here and cutting-edge research is conducted in areas such as:
Several of our researchers are linked to Sweden's single largest research programme; WASP, Wallenberg AI, Autonomous Systems and Software Program, as well as WASP-HS, which enables cutting-edge research in the humanities and social sciences and how artificial intelligence affects people and society.
High-quality education
The Department of Computing Science also offers high-quality and reputable educational programmes, strongly anchored in research, and in partnership with industry. Find out more about the department by clicking one of the links below!
Supervisor Monowar Bhuyan and Professor Erik Elmroth at Sourasekhar Banerjee's nailing ceremony, together with doctoral students and postdocs at one of many events on Monday.