Researchers from ITU receive five grants from Independent Research Fund Denmark
The grants will fund research in artificial Intelligence, urban air and language and other research areas. Five researchers receive the grants for their respective research projects.
Independent Research Fund Denmark (IRFD) grants researchers at the IT University of Copenhagen funding for five research projects. Associate professor Barbara Plank, associate professor Bernardo Machado David, associate professor Steffen Dalsgaard, assistant professor Leon Strømberg-Derczynski, and associate professor Sebastian Risi receive DKK 2,9 million each.
The research projects are within a wide range of research areas under IRFD’s thematic research in Digital Technologies and People and Society.
Machine learning for language understanding
Barbara Plank, associate professor in Natural Language Processing, is among the ITU-researchers who receives a grant from IRFD. Her project aims at developing novel information extraction algorithms for better digital language understanding. Current capabilities to automatically extract valuable information from large text collections are limited, particularly for underprivileged languages such as Danish.
The technology developed through the research project will be evaluated on real use case data provided by the Danish Agency for Labor Market and Recruitment that aims at enabling better job matching and job demand analysis.
Better data security for Smart Contracts
Not until the service has been accomplished, the payment will take place. This is among the advantages of Smart Contracts, a digital contract that process data to ensure that users have honored their parts of the contract.
Through transactions on blockchain based cryptocurrencies smart contracts have become readily available. The challenge is who gets access to watch the transactions in order to meet the GDPR legislation.
Bernardo Machado David, associate professor at Computer Science, receives funding for the research project ’PUMA: Publicly Verifiable Multiparty Computation and Applications’. Through the research project, he will investigate how to build a secure computation protocol for smart contracts that hides private data from other users but allow third parties to verify that the contract has been correctly executed.
He will collaborate with researchers from the university KU Leuven in Belgium, Bar Ilan University in Israel and the blockchain company Concordium in Switzerland and Denmark.
The quality of urban air
Steffen Dalsgaard, associate professor at the Department of Business IT at ITU, likewise receives grants for the research project ’Making Sense of Urban Air’. The project investigates how new and more detailed data on air quality in Copenhagen is made available by Google’s Project Air View, and furthermore what reactions and patterns of behavior the new data give rise to among experts, authorities and citizens.
The project is conducted in cooperation with Copenhagen Solutions Lab, which is a part of Copenhagen Municipality.
Artificial Intelligence in the battle against misinformation and propaganda
Artificial intelligence can make it easier to detect and disprove misleading claims. Preliminary findings at ITU support this. However, misinformation technology is generally restricted to privileged languages like English. The project ‘Verif-AI’ aims to develop technology which can detect misinformation expressed in many languages.
At first assistant professor Leon Strømberg-Derczynski, Department of Computer Science, will lead a team that extracts misinformation data across different languages. Later he will extend models developed by the Natural Language Processing research group at ITU to work for languages that so far have limited or no technology for fact checking and misinformation detection. The goal is to bring automatic and rapid misinformation detection technology to beyond the English-speaking world.
The 30-month project ‘Verif-AI’ will be conducted in co-operation with the EU WeVerify project and with IFCN member TjekDet (‘CheckIt') from the Danish media outlet, Mandag Morgen.
Better generalization in AI
Face recognition, voice recognition, and automatic translations are fields of applications where AI impresses. However, while AI methods perform well on data used for training, they have a limited ability to solve similar problems or varieties of the same problem.
Through the research project ‘QD2L: Improving Generalization in Deep Learning through Quality Diversity’ associate professor Sebastian Risi from the Department of Digital Design suggests a new approach to training AI-methods such as neural networks. The aim is to make them more general and provide further insights into how they work.
Anna Lohmann Ahlbom, Press Officer, phone +45 25 55 04 47, email email@example.com
Bernardo Machado David, Associate Professor, email firstname.lastname@example.org
Leon Derczynski, Assistant Professor, email email@example.com
Steffen Dalsgaard, Associate Professor, phone +45 7218 5251, email firstname.lastname@example.org
Sebastian Risi, Associate Professor, email email@example.com
Barbara Plank, Associate Professor, phone +45 7218 5274, email firstname.lastname@example.org