NWA-ORC awards 1.9 million euros to dr. Arianna Bisazza and consortium for the research project "Opening the Black Box of Deep Learning for Language, Speech and Music"

A total of 21 consortia will work as teams on interdisciplinary research that will bring scientific and societal breakthroughs within reach. In the projects, the entire knowledge chain and societal organisations, both public and private, will work closely together. The projects receive funding in the second round of the Dutch Research Agenda Programme: Research along Routes by Consortia (NWA-ORC).

In total, more than 93 million euros have been made available. The Dutch Research Agenda is contributing 81 million euros, and (international) consortium partners are providing 12 million euros in the form of co-funding.

About the project

Deep learning is producing AI models so accurate that they are significantly affecting individuals, businesses and society. New responsibilities and questions are arising, and the poor interpretability of deep learning is a serious limitation. With the project, titled "Open the Black Box of Deep Learning for Language, Speech and Music", the consortium studies how to make deep learning transparent and “explainable” for speech, text, and music applications.

The Consortium, led by Jelle Zuidema (UvA), consists of the University of Amsterdam, Tilburg University, the University of Groningen, VU Amsterdam, Radboud University, Textkernel, Floodtags, GlobalTextware, TNO, KPN, Deloitte, Algent, Chordify, Stichting Open Spraaktechnologie, Waag, muZIEum, and Huggingface.

About dr. Arianna Bisazza

Dr. Arianna Bisazza is assistant professor in computational linguistics at the Center for Language and Cognition (CLCG) of the University of Groningen and VENI laureate since September 2016. Her research aims to identify the intrinsic limitations of current translation and language modeling paradigms, and to design robust Natural Language Processing algorithms that can adapt to a wide range of world languages.