Helsinki Institute for Information Technology
Helsinki Institute for Information Technology
Heiskanen E.,National Consumer Research Center |
Johnson M.,Helsinki Institute for Information Technology |
Vadovics E.,Green Dependent Sustainable Solutions Association
Journal of Cleaner Production | Year: 2013
European energy policy aims to shift the energy market towards an increased focus on energy services based on end-user needs. This requires a close understanding of the role of end-users and their needs and practices. Based on a European project called CHANGING BEHAVIOUR, we examine the interaction between energy users and energy efficiency practitioners. Using previous cases and our own pilots as data, we uncover the main difficulties in understanding and working with energy users. We argue that formal user research and interaction methods are helpful, yet insufficient for project success or even genuine user responsiveness. Additionally, methods and skills are needed for interacting with broader networks of stakeholders in the user context. Moreover, user responsiveness requires informal interaction with energy users, interpersonal skills and human judgement, which are difficult to develop merely by using better methods. © 2012 Elsevier Ltd. All rights reserved.
Eggeling R.,Martin Luther University of Halle Wittenberg |
Roos T.,Helsinki Institute for Information Technology |
Myllymaki P.,Helsinki Institute for Information Technology |
Grosse I.,Martin Luther University of Halle Wittenberg
Journal of Machine Learning Research | Year: 2014
Inhomogeneous parsimonious Markov models have recently been introduced for modeling symbolic sequences, with a main application being DNA sequence analysis. Structure and parameter learning of these models has been proposed using a Bayesian approach, which entails the practically challenging choice of the prior distribution. Cross validation is a possible way of tuning the prior hyperparameters towards a specific task such as prediction or classification, but it is overly time-consuming. On this account, robust learning methods, which do not require explicit prior specification and - in the absence of prior knowledge - no hyperparameter tuning, are of interest. In this work, we empirically investigate the performance of robust alternatives for structure and parameter learning that extend the practical applicability of inhomogeneous parsimonious Markov models to more complex settings than before.
Heinonen M.,University of Helsinki |
Heinonen M.,Helsinki Institute for Information Technology |
Shen H.,University of Helsinki |
Zamboni N.,ETH Zurich |
And 2 more authors.
Bioinformatics | Year: 2012
Motivation: Metabolite identification from tandem mass spectra is an important problem in metabolomics, underpinning subsequent metabolic modelling and network analysis. Yet, currently this task requires matching the observed spectrum against a database of reference spectra originating from similar equipment and closely matching operating parameters, a condition that is rarely satisfied in public repositories. Furthermore, the computational support for identification of molecules not present in reference databases is lacking. Recent efforts in assembling large public mass spectral databases such as MassBank have opened the door for the development of a new genre of metabolite identification methods. Results: We introduce a novel framework for prediction of molecular characteristics and identification of metabolites from tandem mass spectra using machine learning with the support vector machine. Our approach is to first predict a large set of molecular properties of the unknown metabolite from salient tandem mass spectral signals, and in the second step to use the predicted properties for matching against large molecule databases, such as PubChem. We demonstrate that several molecular properties can be predicted to high accuracy and that they are useful in de novo metabolite identification, where the reference database does not contain any spectra of the same molecule. © The Author 2012. Published by Oxford University Press. All rights reserved.
Shen H.,Aalto University |
Shen H.,Helsinki Institute for Information Technology |
Duhrkop K.,Friedrich - Schiller University of Jena |
Bocker S.,Friedrich - Schiller University of Jena |
And 2 more authors.
Bioinformatics | Year: 2014
Motivation: Metabolite identification from tandem mass spectrometric data is a key task in metabolomics. Various computational methods have been proposed for the identification of metabolites from tandem mass spectra. Fragmentation tree methods explore the space of possible ways in which the metabolite can fragment, and base the metabolite identification on scoring of these fragmentation trees. Machine learning methods have been used to map mass spectra to molecular fingerprints; predicted fingerprints, in turn, can be used to score candidate molecular structures. Results: Here, we combine fragmentation tree computations with kernel-based machine learning to predict molecular fingerprints and identify molecular structures. We introduce a family of kernels capturing the similarity of fragmentation trees, and combine these kernels using recently proposed multiple kernel learning approaches. Experiments on two large reference datasets show that the new methods significantly improve molecular fingerprint prediction accuracy. These improvements result in better metabolite identification, doubling the number of metabolites ranked at the top position of the candidates list. © 2014 The Author. Published by Oxford University Press. All rights reserved.
Lehdonvirta V.,Helsinki Institute for Information Technology
Journal of Technology, Learning, and Assessment | Year: 2010
I argue that much influential scholarship on massively-multiplayer online games and virtual environments (MMO) is based on a dichotomous a ̂creal world vs. virtual worldac model. The roots of this dichotomy can be traced to the magic circle concept in game studies and the cyberspace separatism of early Internet thought. The model manifests on a number of dimensions, including space, identity, social relationships, economy and law. I show a number of problems in the use of this model in research, and propose an alternative perspective based on Anselm Straussa's concept of overlapping social worlds. The world of players does not respect the boundaries of an MMO server, as it frequently flows over to other sites and forums. At the same time, other social worlds, such as families and workplaces, penetrate the site of the MMO and are permanently tangled with the players' world. Research programs that approach MMOs as independent mini-societies are therefore flawed, but there are many other kinds of research that are quite feasible. © 2001-2010 Game Studies.
Pajarinen J.,Aalto University |
Hottinen A.,Asparrow Ltd. |
Peltonen J.,Aalto University |
Peltonen J.,Helsinki Institute for Information Technology
IEEE Transactions on Mobile Computing | Year: 2014
The performance of medium access control (MAC) depends on both spatial locations and traffic patterns of wireless agents. In contrast to conventional MAC policies, we propose a MAC solution that adapts to the prevailing spatial and temporal opportunities. The proposed solution is based on a decentralized partially observable Markov decision process (DEC-POMDP), which is able to handle wireless network dynamics described by a Markov model. A DEC-POMDP takes both sensor noise and partial observations into account, and yields MAC policies that are optimal for the network dynamics model. The DEC-POMDP MAC policies can be optimized for a freely chosen goal, such as maximal throughput or minimal latency, with the same algorithm. We make approximate optimization efficient by exploiting problem structure: the policies are optimized by a factored DEC-POMDP method, yielding highly compact state machine representations for MAC policies. Experiments show that our approach yields higher throughput and lower latency than CSMA/CA based comparison methods adapted to the current wireless network configuration. © 2002-2012 IEEE.
Kuikkaniemi K.,Helsinki Institute for Information Technology
ITS 2013 - Proceedings of the 2013 ACM International Conference on Interactive Tabletops and Surfaces | Year: 2013
My research is a combination constructive design research and practice-led research in the domain of producing novel big screen experiences for school of art and design and department of motion picture, television and production design. As one of the case studies in my thesis research I present a generic user interface for large interactive walls, Kupla UI. Kupla UI applies physics modeled spherical content widgets to present information. It is primarily targeted for multi-user information exploration and holding informal presentations in public spaces, such as exhibitions, commercial spaces and lobbies. Kupla is designed to support multiple simultaneous users, graph-based content hierarchy, flexible installation form-factors, heterogeneous content, and playful interaction. In Kupla design we have developed multiple different states for spherical widgets that differ in terms of visualization, function and physical modeling. These different states help to accommodate different use cases with the same installation. © 2013 Author.
Cui Y.,Nokia Inc. |
Oulasvirta A.,Helsinki Institute for Information Technology |
Ma L.,Helsinki Institute for Information Technology
International Journal of Human-Computer Interaction | Year: 2011
This article explores how people perceive interactive activities in order to inform navigation history design on mobile devices. Following event segmentation method, 12 participants were asked to break 6 episodes of mobile interaction into segments, organize the segments, and identify those deemed representative. Three findings emerged. First, when making sense of mobile interaction, users concentrate on the content objects on which actions are performed. This indicates the value of content-centric designs in navigation history and other mobile user interface designs. The content objects are data objects and their collections meaningful to the person dealing with it, for example, photos, messages, or albums. Second, users tend to employ two-level hierarchies in grouping segments, and use the similarity in content objects and applications as a reference. They deem the segments as representative where objects are created or changed, or where sharing or querying acts take place. These findings indicate how a navigation history design should organize and prioritize mobile interaction events. Finally, event perception shows relatively low interparticipant consensus, which indicates that navigation history designs have to accommodate large individual differences. © Taylor & Francis Group, LLC.
News Article | December 24, 2016
How That Holiday Song Becomes An Inescapable Earworm It's the time of year when seemingly innocent jingles tunnel into my brain. Right now I'm haunted by Ariana Grande's "Santa Tell Me," which I heard for the first time while researching this article. Help me. There may be good reasons why a song like that becomes an earworm, according to Elizabeth Margulis, director of the music cognition lab at the University of Arkansas. Like déjà vu, you might feel like you know the tune, but something throws you off. "[There's] this idea that songs that tend to get stuck are conventional in some ways but also have some little surprising twist," she says. The latest evidence comes from a study published last month in the journal Psychology of Aesthetics, Creativity, and the Arts. The researchers broke down the anatomy of an earworm tune and started looking for melodic features that may lead to compulsive looping. "This was really the first study into this area of melodic features of earworms," says lead author Kelly Jakubowski, a music psychologist at Durham University. "We focused on the pitch and rhythmic elements of a melody." The researchers didn't find earworms marked with anything as specific as a notable chord, note or musical interval. Instead, Jakubowski says, patterns came to light. When the researchers compared songs that 3,000 people deemed either earworms or not-so-catchy, they noticed the earworms were suspiciously lacking in uncommon musical patterns, such as where the melody only rises in pitch without sliding back down, like in the chorus of "Rock'n Me" by the Steve Miller Band. Instead, earworms often had a very common melodic pattern. "They tended to have this general melodic shape, a pattern of ups and downs in terms of pitch," Jakubowski says. And then they surprise us. Something happens in our brains when we hear the same sorts of patterns over and over, says Margulis, who wasn't involved with this study. When we hear a familiar pattern start, our brains start expecting the rest of it. "We're predicting what's about to come next," she says. Jakubowski says songs that obey simple conventions are nice. "The brain is pretty happy to predict what is happening in our [musical] environment," she says. But in her study, earworms usually had a surprising detail. For example, the hit big band song "In the Mood" by Glenn Miller has a familiar up-and-down movement but a lot of uncommonly large shifts in pitch. "It might be some large leap in pitch that's unexpected or more leaps than are expected in your average pop song," she says. "Something unique to add interest to make the brain want to recall or ponder over this melody." Some of the top earworms identified by study participants include Lady Gaga's "Bad Romance;" "Don't Stop Believing," by Journey; and "Moves Like Jagger," by Maroon 5. Those earworms set you up just to pull the rug out from under you. "Once [your brain starts predicting], you can really start to mess with them and do something new and surprising," Margulis says. "Your expectations are violated. There's a norm that is being crossed." Beautiful music tends to do this, too, Margulis says. "People also link those surprising moments to what gives rise to emotion in music or what passages seem expressive," she says. But earworms aren't always beautiful. And great music doesn't always get caught in an uncontrollable loop. Earworms tend to be simple tunes with just enough variation to distinguish them. "Familiar but a little unfamiliar, not too much," says Lassi Liikkanen, a cognitive scientist at the Helsinki Institute for Information Technology. Maybe, Liikkannen hedges, it's our brains way of integrating new, strange variations with the old ones. "Maybe in order for us to pick up more challenging patterns," he says. Holiday songs do have a characteristic, familiar feel. When a songwriter throws a melodic insurrection into the jingle, it may only be a matter of time before the tune spirals maddeningly in a chorus only I can hear (unless I decide to drag everyone my voice can reach into my purgatory with me.) It doesn't help, of course, that every business on the block finds it necessary to drone the same songs all day. The repetition may contribute to triggering earworms, too, Margulis thinks. "You often get this feeling when you listen to music you've heard before that's different than when you listen to music that's new," she says. "There's a link between that feeling of getting swept along and having something get stuck in your head. This is the case where you really are getting swept along. You're doing it in your head and can't stop." According to work from Liikkannen, over 90 percent of people experience earworms weekly. They are often associated with specific times of day or situations, like a song first heard in a morning, or during relaxation. Luckily, there are ways to break the cycle. Jakubowski suggests "engaging with the earworm." Play or sing the song until you feel like it's gone. Or you can distract yourself with another more tolerable earworm, she says. "You can try chewing gum," Margulis suggests. Studies have found this interferes with certain thought processes like memory recall and scanning melodies. And there's always good old-fashioned passive resistance. If you wait long enough, the earworm will subside on its own. It's almost January, after all.
News Article | December 8, 2016
For the first time, information retrieval is possible with the help of EEG interpreted with machine learning In a study conducted by the Helsinki Institute for Information Technology (HIIT) and the Centre of Excellence in Computational Inference (COIN), laboratory test subjects read the introductions of Wikipedia articles of their own choice. During the reading session, the test subjects' EEG was recorded, and the readings were then used to model which key words the subjects found interesting. 'The aim was to study if EEG can be used to identify the words relevant to a test subject, to predict a subject's search intentions and to use this information to recommend new relevant and interesting documents to the subject. There are millions of documents in the English Wikipedia, so the recommendation accuracy was studied against this vast but controllable corpus', says HIIT researcher Tuukka Ruotsalo. Due to the noise in brain signals, machine learning was used for modelling, so that relevance and interest could be identified by learning the EEG responses. With the help of machine learning methods, it was possible to identify informative words, so they were also useful in the information retrieval application. 'Information overload is a part of everyday life, and it is impossible to react to all the information we see. And according to this study, we don't need to; EEG responses measured from brain signals can be used to predict a user's reactions and intent', tells HIIT researcher Manuel Eugster. Based on the study, brain signals could be used to successfully predict other Wikipedia content that would interest the user. 'Applying the method in real information retrieval situations seems promising based on the research findings. Nowadays, we use a lot of our working time searching for information, and there is much room in making knowledge work more effective, but practical applications still need more work. The main goal of this study was to show that this kind of new thing was possible in the first place', tells Professor at the Department of Computer Science and Director of COIN Samuel Kaski. 'It is possible that, in the future, EEG sensors can be worn comfortably. This way, machines could assist humans by automatically observing, marking and gathering relevant information by monitoring EEG responses', adds Ruotsalo. The study was carried out in cooperation by the Helsinki Institute for Information Technology (HIIT), which is jointly run by Aalto University and the University of Helsinki, and the Centre of Excellence in Computational Inference (COIN). The study has been funded by the EU, the Academy of Finland as a part of the COIN study on machine learning and advanced interfaces, and the Revolution of Knowledge Work project by Tekes. Video: https:/ HIIT augmented research http://augmentedresearch. Tekes Re:Know https:/ EU:n MindSee project http://mindsee. Department of Computer Science http://cs. HIIT http://www. The Centre of Excellence in Computational Inference (COIN) http://research.