HIIT supported the Helsinki ICT Community by providing funding for the following activities:
- Organising events such as workshops, conferences, hackathons, bootcamps, and summer schools.
- Initiating new collaboration on potentially high-impact research challenges, often in a multi-disciplinary and cross-university setting.
- Inviting postdoctoral researchers for short visits that may lead to recruitment.
- Visiting high-profile universities or other organisations abroad.
- Boosting the careers of young researchers by providing short-term “bridge funding”.
This funding was not limited to the HIIT research programmes, but was open to the whole Helsinki ICT community. In addition, it served an important purpose by being applicable to cases where other funding instruments were not easily available or the funding from other sources was not sufficient.
For a sample of the activities of the HIIT programmes, please see Highlights.
In 2019, 28 Community Support proposals received about 237 000 euros in total. 80% of this was spent on nurturing new collaboration and 16% on supporting events. In addition, HIIT financed four research visits from Finland abroad. No applications in the other categories (inviting postdoc candidates to Finland, boosting the careers of young researchers) were received this year.
The largest investments in new collaboration were directed to starting joint research projects with The Alan Turing Institute, utilising AI and computational creativity methods in media production, and exploring the possibilities of VR and smart textiles in physical therapy. The Fulbright-Nokia Distinguished Chair in Information and Communications Technologies grant of Professor Brendan Mumey was supplemented with HIIT funding to enable a six month research visit at the University of Helsinki, and “seed funding” was provided to multi-disciplinary projects ranging from automatic detection and classification of marine plastics to Animal-Computer Interaction.
Among the supported events there were the international conferences MyData, RIOT Summit and ICWE, smaller workshops combining computing with astronomy, social sciences and sustainability science, as well as the hands-on programming sprints Helsinki Digital Humanities Hackathon and HIIT Open programming contest. Machine Learning Coffee Seminar continued as a weekly event with the location alternating between Otaniemi and Kumpula.
Examples of supported activities
Multidisciplinary pilot project explored smart textiles and virtual reality in rehabilitation of stroke patients
The mc² – Mobile Cloud Computing group at Aalto University is doing experimental systems research on Virtual Reality (VR) and wearable sensing technologies. In a recent project with Kuopio University Hospital and Savonia University of Applied Science, they explored physical therapy as a potential application area. The idea was to create VR exergames and accompanying accessories for rehabilitation of stroke patients.
Members of the mc² group organised multiple workshops with rehabilitation professionals, in which they developed ideas for possible exercises and acquired feedback on prototypes. Physiotherapy students at Savonia also conducted a survey on the physical exercises used in occupational stroke rehabilitation and carried out an early user test with the exergame system developed in the project.
Two major prototype devices were built: a smart glove and a smart cushion. The glove measures finger bending and gripping force using velostat sensors while the cushion measures the leaning direction and degree of a sitting person. Both devices saw multiple prototype iterations with continuous improvement. The used sensor structures consist of soft materials with intermediate conducting wires leading to a copper wire connection going to the microcontroller. The VR exergame system was built using Unreal Engine 4.
The glove was utilised e.g. in an exercise in which the patient needs to reach out for ingredients while baking in virtual reality. An example of an exercise with the cushion involves shifting the weight in a sitting position to balance a raft floating across the sea. These and some other exercises are described in more detail in a Master’s thesis done in the mc² group.
The team has received funding from Business Finland to create a commercially viable VR exergame system for stroke patients. The experience and connections gained from the early system iterations and close collaboration with rehabilitation professionals laid a strong foundation to move the project to the next level. In addition to Savonia and Kuopio University Hospital, the mc² group has also started active collaboration with Helsinki University Hospital, which rehabilitates the largest number of stroke patients in Finland.
Momentum of the MyData movement continues
The conferences of MyData Global have established their position as the premier events for the rapidly growing community interested in human-centered personal data management. MyData 2019, organised on 25-27 September in Helsinki, brought together 975 participants from 46 countries. The programme was arranged in 9 thematic tracks with “Rebuilding trust – for human-centered data economy” as the overarching theme.
The core idea of MyData is that individuals should be in control of their own data. The MyData approach aims at strengthening digital human rights while opening new opportunities for businesses to develop innovative personal data based services built on mutual trust. More information on the principles and goals advocated by the movement can be found in the MyData Declaration.
The annual conference was organised for the 4th time, and the number and diversity of the participants has been growing steadily. In 2020, there will be 4 conferences on 4 different continents instead of a single global event. Continuous activity is also taking place around the world in MyData Local Hubs and Thematic Groups.
HIIT played a major role in the inception of the MyData movement and is proud of its continuing success. Interesting opportunities for further collaboration may exist especially within the framework of FCAI and HiDATA, both of which recognise privacy and trust as vitally important elements of data intensive research and technology development.
Automatic dialogue generation demonstrated in an international conference on games
How does natural language generation relate to games? These two concepts that are seemingly far away from each other are surprisingly related. In the 14th International Conference on the Foundations of Digital Games (FDG 2019), two doctoral students of the University of Helsinki, Khalid Alnajjar from computer science and Mika Hämäläinen from language technology, presented their work on creative dialogue generation for the Fallout 4 role-playing game.
Alnajjar’s PhD topic focuses on computational linguistic creativity, where he researches computational methods for interpreting and generating figurative language. Computational creativity is a subfield of artificial intelligence that studies human-like creative tasks from a computational point of view. A vision of computational creativity is to produce systems that support our creativity and help us be more creative.
“In my previous work, I have usually worked with metaphors and short expressions such as slogans and headlines. Conducting and learning about natural language generation research in the field of digital games was not only new and interesting to me but it also enriched my knowledge about how natural language processing and natural language generation methods can be employed extensively in games – from teaching players a language to producing dialogues and analysing games for suggesting desired games, and more – to provide players with a better experience”, Alnajjar explains.
The demonstration of the automatic dialogue generation method caught the interest of many attendees. The generated dialogues were dynamically adapted to the game state, and while they often sparked laughter in the audience, most of the attendees were impressed by the modified version of the game. Usually the approach and the output are the most important aspects to consider in research on natural language processing and natural language generation. It appears that the practicality and agility of the method are as crucial in games.
Foundations of Digital Games is a major international conference that covers a broad range of topics, including artificial intelligence for games, modelling players, virtual reality, and theories on games and designs. In 2019, the conference was organized in San Luis Obispo, California, USA.
Links to the articles:
- K. Alnajjar & M. Hämäläinen: A Creative Dialog Generator for Fallout 4
- M. Hämäläinen & K. Alnajjar: Creative Contextual Dialog Adaptation in an Open World RPG
Foundations of animal-computer interaction explored with monkeys of the Helsinki Zoo
Much like humans, animals have also been using computers for a long time. Historically animals have used computer technology in space, used lexigrams (symbol keyboards) to form language, and dolphins have used underwater keyboards. Drawing parallels to human-computer interaction (HCI) research, which studies how humans use computers, ACI aims to investigate how animals interact with technology and the design of related devices. These technologies are developed to be used with animals in zoos and sanctuaries, working animals, domesticated animals in our homes, and wild animals. In HCI, user-centred design is an established methodology for designing with the users by including them as participants within the design process itself.
As computer systems are becoming more and more widespread, Roosa Piitulainen, Ilyena Hirskyj-Douglas and Tapio Takala from Aalto University Computer Science department have recently started looking at how technologies can be used to support the welfare of animals housed in the zoo environment. The work is done in collaboration with the Helsinki Zoo, and applies user-centred design and playful interaction to investigate how we can design computer systems for these animals.
Of particular interest is exploring what it means to have animal-controlled technologies, and how we can support the animals’ autonomy over the technologies that they use. For example, the animals could use the technology for changing their environment or interacting with visitors.
Part of this question, and the starting point for product specifics, is what should the zoo-technology do and how should it be used? To explore what the requirements are in designing and building zoo-technological systems, the team is also researching how requirements should be formed and gathered from the various perspectives of the people who interact with the animals and from the animals themselves. These requirements are then being used for building devices for white-faced sakis in the Amazonian House of the Helsinki Zoo.
For more information and updates, please see https://zoo.cs.aalto.fi/
Siamese neural networks applied to algorithm selection in recommender systems
The performance of a recommendation algorithm varies in different scenarios, and intuitions about what algorithms are best suited to a given scenario can be elusive even to recommender system experts. Furthermore, choosing a single algorithm that performs well across all the scenarios often reduces the effectiveness of the system in specific scenarios. In the domain of scholarly article recommendation, for example, an online evaluation of 33.5M recommendations delivered across multiple applications was performed and it was found that the best performing algorithm in one application (Document embeddings; Click-through rate (CTR): 0.21%) was the worst performing in another (CTR: 0.02%).
Avleen Malhi, a postdoctoral researcher at Aalto University, explored Siamese neural networks (SNN) as a potential solution to this problem during her research visit to Trinity College Dublin, Ireland. Besides contributing to the research field in general, the objective was to improve the recommendations-as-a-service system for digital libraries Mr. DLib.
A Siamese neural network is an artificial neural network, which is typically used for comparing similar instances in different type sets. Siamese neural network selection aims to identify the group of similar algorithms to use for all the instances in a given situation. It does this by calculating the relative absolute error for all algorithms on a particular instance and then ranking the algorithms based on their performance. Finally, relative performance is calculated compared with the best algorithm, which enables the similar performing algorithms to be identified. The key idea is that if an algorithm is the best for a particular instance, the same algorithm can be used for recommendation with similar instances later. Siamese neural networks were applied to algorithm selection on the MovieLens dataset and on scholarly article recommendation. The aim was to use this approach to select the best algorithm for each data instance, and for each recommendation request received.
Avleen’s main collaborator during her 4-week research visit was Joeran Beel, Assistant Professor in Intelligent Systems at the School of Computer Science of Trinity College. Professor Beel is also affiliated with the ADAPT research centre, which is a dynamic research centre in Dublin with experts producing ground-breaking digital content innovations. The visit paved the way for further collaboration and networking within the field of intelligent systems. The next major step will be the submission of a research article in ACM conference RecSys 2020, which is to be held in September next year.
Collaboration with Novartis in the field of Bayesian modelling
With support from HIIT, Topi Paananen from the Probabilistic Machine Learning (PML) group at Aalto University, recently did a 3-month research visit to the pharmaceutical company Novartis in Switzerland. During the visit, Topi and his collaborators worked on methods for constructing prior distributions in Bayesian analyses on longitudinal data sets. The methods were used with real medical data from a longitudinal study and were implemented with the statistical computation tool Stan. Collaboration with Novartis continues after the research visit with the goal of publishing a joint research article.
Thermal cameras hold promise for improved energy management in smart buildings
Improving energy management of buildings is among the most promising application areas of new technology for sustainable living. The Pervasive Data Science Group of the University of Helsinki is experimenting with thermal cameras to gather fine-grained information for use as actionable feedback to heating, cooling and ventilation (HVAC) devices in shared living spaces. The idea is to develop solutions that go beyond reducing energy usage and can balance between living quality and overall energy consumption.
As a first step in this direction, they have conducted a feasibility study about using thermal imaging to estimate changes in comfort. Climate factors, such as exposure to sunlight, can influence thermal sensation and personal comfort level in a room by affecting the extent of thermal radiation seeping through into the room. To assess whether thermal imaging can indeed capture these factors, an office room shared by four employees and occasional visitors was monitored for three weeks. A Caterpillar CAT S60 phone with an integrated FLIR thermal camera was set up to take thermal images every four minutes, and temperature statistics from the images were compared to data provided by the Finnish Meteorological Institute (FMI). The results were promising, demonstrating how the influence of external climate factors, such as sunlight, can be captured using thermal imaging. In particular, a significant correlation was found between cloud cover statistics and temperature statistics extracted from thermal imaging output.
To understand how thermal imaging measurements correlate with the occupants’ perceptions, as part of the feasibility study the participants were asked to fill in daily surveys about their level of comfort and factors that could influence it. The preferred temperature is a subjective metric that varies from one person to another. Seven-point Likert scales were used for both thermal sensation (“very cold” – “very hot”) and comfort (“highly uncomfortable” – “highly comfortable”). The participants were also asked about the clothes they were wearing, how long they spent in the room, whether any windows or doors were open, and whether there were other events that may have affected their comfort. The results showed that comfort varied across the participants and that different areas of the room were exposed to differing amounts of thermal radiation, motivating the need for solutions that can separately analyze factors affecting thermal comfort of individuals in shared spaces.
Motivated by the promising results of the feasibility study, the team is continuing research into developing methods for assessing thermal comfort using thermal imaging. The primary focus is currently on developing methods for separating people from the thermal images and characterizing thermal radiation for different regions of the space being monitored. The feasibility study was supported by HIIT and the work is continuing through other funding sources.
Aalto’s Data Management Group develops cross-disciplinary collaboration in computational social science
Marta Kołczyńska was a visiting researcher at the Department of Computer Science of Aalto University from 21 April to 23 June, 2019. Marta is a post-doctoral researcher in sociology at the Institute of Philosophy and Sociology of the Polish Academy of Sciences. During the visit, she collaborated with the group of Aristides Gionis on applying computer science tools to social science data, which crystallized into two problems.
The first deals with the ways of measuring social polarization using survey data measuring political attitudes, such as support for the welfare state or immigration, or trust in state institutions. Social polarization refers to the extent to which groups of individuals are divided and distant from each other with regard to certain characteristics, such as economic status or political attitudes. While increases in social polarization around ideological issues or policy preferences are considered one of the threats facing contemporary democracies, the methods of measuring polarization of attitudes have thus far been rather limited. During the collaboration the team explored the application of clustering techniques and feature selection to cross-national survey data with measures of political attitudes from the European Social Survey, and is continuing to work on devising polarization measures to capture the societal dynamics that European societies undergo.
The second problem deals with the role of status in digital communities. Wikipedia is unique in that it is manifestly meritocratic and status is determined solely by the value of the contribution to the community in the form of improving entries, reacting to vandalism, and upholding community values. The team collected data on promotions to adminship on English language Wikipedia as well as characteristics of the candidate and the voters since the initiation of the process in 2003. These data will be used to analyze the role of different status dimensions in the Request for Adminship process.
Research collaboration with Stockholm University in the field of Human-Computer Interaction
With the aid of HIIT Community Support, Jesse Haapoja, who is part of the Digital Content Communities (DCC) group at Aalto University’s Department of Computer Science, visited the Post-Interaction Computing group (PIC) at the Department of Computer and Systems Sciences at Stockholm University, Sweden. The visit lasted for three months during the spring of 2019. PIC is a multidisciplinary research group that focuses on Human-Computer Interaction. It is led by Professor Barry Brown. The group’s recent research has focused on mobility and mobile technologies, interpersonal and economic encounters, and using speech and gaze as design materials for human-computer interaction.
During his stay, Haapoja worked with HIIT-alumni Dr. Airi Lampinen, who is an Associate Senior Lecturer at the Department and one of the three faculty members in the PIC group. The visit was a part of Haapoja’s doctoral studies and related to the Kone Foundation funded project Algorithmic systems, Power and Interaction for which Lampinen acts as the principal investigator. The project studies how people interact with and resist algorithmic systems and what kind of power these systems have in the society, and how this power should be conceptualized and studied.
In addition to familiarizing himself more with the type of HCI research and methodologies that the PIC group specializes in, Haapoja worked on two specific projects during his stay. First, related to the Algorithmic systems project, Haapoja finalized and submitted to peer-review a manuscript co-authored by Lampinen and Dr. Salla-Maaria Laaksonen from the Centre for Consumer Society Research on gaming automated hate-speech detection. The article studied a project where municipal election candidates’ social media messages were automatically monitored for hate-speech and negative reactions that the project incited. Second, together with Associate Professor Rob Comber from the Royal Institute of Technology KTH and Lampinen, Haapoja worked on a project to envision what human-computer interaction could be beyond the interaction, that is, what might be left unnoticed if research only focuses on those aspects of interaction that are intentional from the user’s or designer’s perspective and how HCI might better account for such issues in both empirical studies and design. This collaboration has also led to an article submitted to peer-review.
Finally, during the visit, Haapoja co-organized the first workshop of the NOS-HS funded series of meetings on Nordic perspectives on Algorithmic systems. This multidisciplinary workshop series brings together researchers from Finland, Sweden, and Denmark to discuss what could be a Nordic way of approaching different kinds of algorithmic systems as they gain constantly more importance in our daily lives.
Collaboration with Stockholm University continues with the still ongoing Algorithmic systems project and the workshop series.
HIIT Open 2019 programming contest
The 4th HIIT Open programming contest was organised on 25 May 2019 in Otaniemi. The contest was open to everyone interested in programming and algorithmic challenges, and in addition to the universities in the Helsinki region, there were participants from Finnish high schools and companies, and two teams from Tallinn University of Technology. The youngest team, named “Piltit”, was not even in high school yet. The total number of contestants was 32.
The 14 teams with 1-3 members in each had 13 tasks to solve, and 5 hours of time. In each task they had to write a computer program that solves a given task correctly and efficiently.
The winning team “bigint bugaa”, with Roope Salmi (Olarin Lukio), Olli Järviniemi (Päivölän opisto) and Antti Röyskö (University of Helsinki), solved 11/13 tasks correctly. The team “Game of Nolife”, with Tuukka Korhonen (University of Helsinki), Olli Hirviniemi (University of Helsinki) and Otte Heinävaara (Princeton University), had the same number of correct solutions, but came second because they submitted their solutions later. The 3rd place went to the team “Varokaa J:tä”, with Siiri Kuoppala, Henrik Lievonen and Miska Kananen, all from Aalto University.
For more information on the HIIT Open programming contest, please see http://open.hiit.fi/ – the contest has been organised since 2016 in collaboration between HIIT, Aalto University, and the University of Helsinki.
Helsinki Digital Humanities Hackathon #DHH19 | 15.-24.5.2019
Helsinki Digital Humanities Hackathon #DHH19 gathered students and researchers of humanities, social sciences, and computer science in May at the University of Helsinki. During a week and a half of intensive multi-disciplinary work, the groups applied digital methods to a variety of datasets, with the goal of solving research questions in the following themes:
- The Many Voices of European Parliamentary Debates
- Genre and Style in Early Modern Publications
- Brexit in Transnational Social Media
- Newspapers and Capitalism
This year, the event was organized as an international summer school sponsored by CLARIN – European research infrastructure for language resources and technology, and DARIAH – digital infrastructure for arts and humanities, attracting participants from 14 different countries.
“DHH is a concept that we have been developing for more than 5 years. Last year we mixed international students with local University of Helsinki and Aalto students for the first time. We were quite curious how this will work out because not only are we attempting a project course that is truly multidisciplinary but also international. This year we proved that the idea of going international was fruitful. In 2019 we had 4 excellent groups with interesting results. We are also grateful to CLARIN and DARIAH for making this possible. See you next year!”, comments Mikko Tolonen, one of the organizers of the Digital Humanities Hackathon.
Read the blogs of the Hackathon groups: https://blogs.helsinki.fi/digital-humanities-hackathon/
More information about DHH19, including the groups’ final presentations – video recording, slides, and posters: http://heldig.fi/dhh19
Application of big data methods in sustainability science discussed in a workshop
Big data and increasing computing power can contribute to more sophisticated monitoring and modeling to address sustainability challenges and enable more informed decision-making. For instance, ecological monitoring towards resource optimisation, environmental risk assessment, and real-time reporting and assessment of environmental quality indicators are domains where big data is increasingly used. At the same time, environmental social scientists are discovering the potential of text data mining to gain insights in framing and narrative aspects of sustainability transitions.
Much of the work is still at an experimental stage, however, and there are no detailed guidelines for applying these methods and understanding their limitations. As part of the spring 2019 Sustainability Science Days, Daria Gritsenko, Polina Rozenshtein, Dalia D’Amato and Karoliina Isoaho organised a workshop titled Big data methods: what is the contribution to Sustainability Science? The idea was to invite scholars and practitioners to inquire and discuss the validity, meaning, policy relevance and governance consequences of big data methods in sustainability science.
The invited guest speaker in the event was Dr. Sanja Šćepanović from the Social Dynamics team of Nokia Bell Labs in Cambridge. She talked about two lines of her past and current research that tackle sustainability challenges from different angles. On the one hand, she is interested in individual behavioural change and involved in a project where teams from several EU universities are investigating how to involve and encourage citizens for more efficient energy consumption. On the other hand, increasingly available satellite imagery facilitates understanding of larger scale phenomena and can support sustainable urban development.
The thematic focus of the Sustainability Science Days 2019 was on sustainable production and consumption. During two days this challenge was approached via scientific debate and popular discussion events. The conference was organised jointly by Helsinki Institute of Sustainability Science (HELSUS) and Aalto Sustainability Hub (ASH).
5G enabled IoT innovations created at IoThon
5G mobile networks are tested and built all around the world and companies are looking for new business areas and service concepts enabled by 5G. IoThon 2019 brought together 110 students and developers from all over Europe for a 48-hour hackathon with challenges set by Ericsson, Nokia, Siemens, Telia and RIOT community. The challenges ranged from novel use cases for 5G light poles to improving the environment of the Otaniemi campus area.
Aalto University and the industrial partners provided platforms that enabled access to the internals of the Otaniemi campus 5G mobile network. The participants could deploy sensors in any location within Otaniemi and implement their own services based on the data from the sensors.
The most popular challenge was set by Nokia: 17 out of 27 teams developed new digital services and business opportunities for smart LuxTurrim5G light poles. The solutions included for example using light poles for charging electric scooters, finding lost pets and warning drivers of a near-by deer.
The COMNET main prize was awarded to team Quadratic that took part in the Telia challenge. They developed an augmented reality coach that gives athletes real-time haptic feedback through actuators.
“The winning team produced a working prototype combining motion capture from mobile phone video feed and haptic feedback through actuators. The demonstrated system made use of the 5G low-latency communications and novel machine learning tools to provide a novel service that would be very beneficial for sports training. Especially, the jury appreciated the level of system integration the team managed to do in the given short time.”, explained the chairman of the jury, professor Riku Jäntti.
”We are happy that we found a solution that uses 5G network and that we got it working during the weekend,” said Perttu Yli-Opas from the Quadratic team. ”We most probably don’t have the time to take our solution any further. If someone with coding skills would like to continue with the solution there is potential to even take it to the market. Telia awarded us with a possibility to demonstrate the solution at ITS Europe Conference 2019 in Helsinki in June. This would bring great visibility for the sports coach!”
Nordita program on Solar Helicities brought together observers, modelers and theorists
Understanding the Sun to the extent that we can predict how its magnetic field drives space weather and climate is an extremely challenging but important problem for the present-day, increasingly hi-tech, society. Key variables in this process are solar helicities, broadly speaking meaning twist of velocity or magnetic field. Magnetic helicity, for example, can be linked to the eruptibility of active regions. Research on helicities is a combination of observational efforts dealing with massive amounts of data, computational research building models on how the magnetic fields are generated and transformed into helical active regions, and theoretical studies of magnetic helicity and reconnection.
The Solar helicities program was organised in Stockholm, NORDITA, in March 2019. The first week of the month-long program took the form of a kick-off focus event, bringing together nearly 60 participants. The work continued for three weeks more in dedicated working groups, on topics formulated during the focus event. These topics included the role and importance and evolution of helicities in natural systems, observations and measurements of helicities, magnetic dynamos, their connection to helicities, and their modeling, and helicities and space weather.
With HIIT support, computer science and space technology students could participate in this work, providing them a unique opportunity to apply their methods on scientifically and societally significant data sets, and at the same time obtain guidance from the international experts that design and conduct the space missions and those that carry out the modeling tasks.