5
Research
Research is one of NORA’s core pillars. As laid out in the strategic plan, NORA aims to be an internationally known research and education network.

Norway has a great potential to succeed in the field of AI research. To utilise this potential, NORA is working to offer plat­ forms for both fundamental and applied research that reflects the complexity, inter­-disciplinarity and diversity in the field of AI.

NORA has prioritised establishing and strengthening the NORA Research School as a research and education platform for NORA partners. Read more about the NORA Research School in Chapter 6.

International cooperation is fundamen­tal in bringing Norwegian AI research to the forefront and to make NORA partners’ research internationally visible. To give exposure to the Norwegian researchers abroad, NORA has entered into agree­ments with renowned institutes like the Alan Turing Institute in UK and the Helmholtz Information & Data Science Academy in Germany. The agreements pave the way for research residencies at the institutes for Norwegian researchers. Read more about international coope­ration in Chapter 9.

NORA has also taken steps to actively engage and coordinate Norwegian Cent­res for Research­based Innovation (SFIs) where NORA partners are involved, like Visual Intelligence (UiT), Big Insight (UiO) and Media Futures, with regular meetings. NORA has integrated input from these stakeholders in initiatives like the NORA Research School. Furthermore, NORA has also actively connected researchers for relevant calls for funding proposals, consortium building and cooperation.

NORA has organised many research webinars where various researchers at NORA partners have been engaged in sharing their research with the wider community. NORA has also organised workshops on cutting­ edge topics and was one of the first initiators in the world to organise a workshop discussing the recent advances in methods of protein folding through the algorithms AlphaFold and RosettaFold.


5.1
Connecting partners: AI-Mind

The AI­-Mind application was successful, and 2021 was kick­off for this 14 million Euro project, the largest Horizon Europe R&I project within AI led from Norway
AI­-Mind is a five­-year EU­-funded Horizon 2020 project on screening brain connec­tivity and dementia risk estimation in peo­ple affected by Mild Cognitive Impairment (MCI). AI­-Mind is developing two AI­-based tools that will identify dysfunctional brain networks and will assess dementia risk: the AI­Mind Connector and the AI­-Mind Predictor. The AI­-Mind Connector will fully automate the identification of early brain network disturbances; after enriching data from AI­-Mind Connector with genetic and cognitive information, AI­-Mind Predictor will provide an early marker of risk for de­mentia in people with MCI. Thus, the aim is to equip healthcare professionals with innovative tools that will enable timely diagnosis and extend the window for preventive interventions and therapies.

NORA has been supporting AI­-Mind and the project coordinator, Dr. Ira Haraldsen, through the application phase. Through NORA, Dr. Haraldsen has connected with AI researchers at OsloMet and to inter­national researchers through CLAIRE.
The AI­-Mind application was successful, and 2021 was the kick off for this 14 million Euro project, the largest Horizon Europe R&I project within AI led from Norway.

The project’s consortium comprises medical experts and opinion leaders on dementia, experts on AI, brain signal ana­ysis, and computer science, SMEs and academic spin­-off companies, patient and professional stakeholders and health technology assessment experts.

NORA now offer guidance and advice in the critical aspects of the AI­Mind project. We contribute as external support to all tasks related to AI­modelling of the AI­-Mind Connector and AI-Mind Predictor. NORA’s CEO Klas Pettersen is an active member of AI­-Mind’s Scientific Advisory Board. Among others, Pettersen partici­ pated in person in the project’s General Assembly, which was held in Oslo in Sep­tember 2021.

5.2
Journal: Nordic Machine Intelligence

“2021 was the grand start for the NMI journal. We had a pro­ductive year publishing high­quality conference papers and are looking forward to an even more productive 2022 with excellent conference papers and journal papers”.
– NMI’s Editor-in-Chief, Anne Håkonsson
To promote open science and create a Nordic community within AI, NORA has launched the pan­ Nordic journal Nordic Machine Intelligence (NMI). The journal will publish a wide range of articles related to promoting research and education in all aspects of AI. The first volume of NMI was published on 1 November 2021 and published the results from the MedAI: Transparency in Medical Image Segmen­tation challenge.

The aim of the NMI journal is to provide a high­quality journal with complete, ac­curate, and concise research papers of interest for the international public arena. The ultimate goal is to position the NMI journal on the second level Norwegian Scientific Index bibliographic database.
NMI is a collaborative project with Anne Håkansson (UiT) as NMI’s Editor­ in ­Chief. Morten Goodwin (UiA), Klas Pettersen (NORA) and Michael Riegler (Simula­-Met) are the journal’s Associate Editors. Bjørn­Jostein Singstad at Oslo University Hospital is the journal’s General Manager.

NMI is published through FRITT ­A publication service at University of Oslo (UiO) for researchers and groups who want to establish a new scholarly Open Access journal or to convert an existing journal to Open Access.

5.3
EU Network for AI

“NORA.EU will undoubtedly be able to facilitate more and better applications for Horizon Europe from Norwegian researchers. NORA.EU will also be able to promote Norway’s interests in a field in rapid development”.
– NORA’s CEO Klas H. Pettersen
NORA was granted funding by the Research Council of Norway to build an AI focused EU Network for Norwegian stakeholders.
The network, called NORA.EU, will help to mobilise and support researchers in the fields of AI to come together to apply for funds from Horizon Europe.

NORA has, since its inception, worked towards bringing Norwegian AI research­ers together alongside other societal and industrial actors. By becoming an EU­ net­work for Horizon Europe, NORA is now amplifying the Horizon Europe calls for Norwegian participation, facilitating the creation of consortiums and supporting relevant applicants through workshops, networking events and focused training programmes and activities, which will increase Norwegian participation in the Horizon Europe framework.
Internationally, NORA.EU has a close co­ operation with several institutions, among others with CLAIRE. NORA hosts CLAIRE’s office for the Nordic countries and the UK. NORA.EU has a steering group with broad representation from NORA partners, Inno­vation Norway and the Research Council of Norway. It also has several forums. The network is co­lead by NORA ́s CEO along with Digital Norway CEO Liv Dingsør.

5.4
Data Competition

Michael Riegler (OsloMet)
Organizer and driving force behind the dataset competition
Competitions have been important for creating progress in the field of AI. The ImageNet* competition had an impact on creating the field of Deep Learning, and more recently the CASP** challenge has produced groundbreaking algorithms such as AlphaFold and RosettaFold. Building on the principles of open innovation and transparent science, NORA believes that data competitions are a key factor in en­abling collaboration as well as competition among researchers. This may further result in creating the desired impact and visibility for Norwegian research in the field of AI.

During the summer of 2021, NORA, in collaboration with Simula Research Lab­oratory, launched a dataset competition called MedAI: Transparency in Medical Image Segmentation. NORA and Simula proposed a task that focused on medical image segmentation and transparency in machine learning­ based systems.

Three tasks to meet specific gastrointestinal image segmentation challenges collect­ed from experts within the field were proposed. This included two different segmentation scenarios and a task on transparent machine learning systems that emphasises the need for explaina­ble and interpretable machine learning algorithms. Results of the competition were announced at the NordicAIMEET conference. The winners were Adrian Galdran and Debayan Bhattacharya. NORA is grateful to Michael Riegler and his team for his extraordinary efforts in developing the competition and reviewing submissions.

* ImageNet (image­net.org)
** CASP – Wikipedia