Research highlights

---

CityWall multi-touch display

Installed in downtown Helsinki, CityWall is a multi-touch display featuring digital media arranged into themes and events. Citizens can shuffle the content as if the images are real, creating a customisable digital experience and promoting meaningful digital media interaction. CityWall was created by the Ubiquitous Interaction research group in the IPCity project.

---

Mobility and cognition

The long-term objective of this line of research is to understand qualitatively and quantitatively the impact of mobile computing and communication to the interactive behaviour of users and user groups. To this end, the research has focused on three major lines: 1) the investigation of cognitive regulation of action in mobile human-computer interaction; 2) the description of the fundamental limitations in interacting with mobile devices when mobile; and 3) the charting of possible user interface solutions.

During the research, several innovative research methods and instruments have been developed to facilitate experimental research in naturalistic real-world settings. For instance, HIIT has developed a state-of-the art wearable video recording system that makes it possible to collect rich data for mobile human-computer interaction studies.

Availability of such data has enabled us to study phenomena that would not appear in a laboratory setting. As an example, we built a predictive model of a mobile user’s attention, basing on Bayesian networks and data collected from 28 users of mobile web browsers. The results are promising, with accuracy in binary classification reaching 72% (22% above default), even with realistic sensors.

---

Social media, especially mobile photography and mobile spectator media

In this line of work, mobile media services, especially for social photography and large-scale events, have been conceptualized, developed and extensively tested. This work has resulted in service design principles for mobile group media, as well as explorative implementations in commercial products (Kuvaboxi, Jaiku, Comeks), service prototypes (Comedia), and open mobile application platforms (MUPE). The work has been performed in close co-operation with UC Berkeley (prof. Marc Davis and prof. Nancy van House).

---

Techniques and tools for learning linear latent variable models

A common data-analysis framework for continuous data is to describe the data as a linear mixture of some underlying hidden variables. This family of methods includes Independent Component Analysis (ICA) and Non-negative Matrix Factorization (NMF), which both have received considerable attention in the machine learning community. We have contributed significantly to the problem formulation, solution algorithms, and software for these methods. In particular, we have published a book which is now the standard reference on ICA (Hyvärinen et al., Independent Component Analysis, Wiley, 2001). We have also developed and improved the FastICA MATLAB package, implementing the world-wide most widely-used ICA algorithm, which we developed in the 1990’s.

Furthermore, we have focused on the important problem of estimating the reliability of ICA components (Himberg et al., NeuroImage, 2004). We have extended the standard NMF method to include sparseness constraints. The resulting method (Hoyer, JMLR, 2004) has become a main reference for modern approaches to NMF, and our corresponding MATLAB package is widely-used.

---

Finding orders from data

In certain data analysis applications there is a natural ordering for the rows or columns of the data. For example, in paleontological presence/absence data the rows represent sites and the columns represent species: the task is to find an ordering for the sites so that for each species its occurrences are in consecutive observations. In the error-free case this seriation problem reduces to consecutive ones problem, but it is NP-hard for realistic data. We have in the last years developed novel algorithms for this seriation task (Gionis et al., Paleobiology 2006; Puolamäki et al., PLoS Computational Biology 2006); their performance is excellent compared to previous approaches. Recent results (Gionis et al., KDD 2006) show also that finding partial orders can be done efficiently.