Home » Events 2009

Events 2009

27 Nov 10:00 Workshop on Multitouch: Design Issues and Knowledges: Limitiations and Affordances

When: Friday 27.11.2009 10am- 1pm.
Where: Place: Pluto cabinet at Pilotti Building, 5th floor, Spektri

We have an international visitor working with Multitouch and other interesting novel applications and have arranged an informal workshop for 10am -1pm noon on Friday 27th November with one 45 minute presentation and 4 X 15 minute presentations and discussions.

Our visiting speaker is Hyowon Lee from Dublin City University
http://www.compapp.dcu.ie/~hlee/

1) Hyowon Lee: Designing Novel Applications Inspired by Emerging Media Technologies will show some different multimedia applications on a variety of platforms that he has designed, and looks to raise design issues for the ubiquitous use of different interaction platforms. He asks how can we leverage this design knowledge that is currently growing for different platforms/devices?

Hyowon will give a 40 -45 minute talk (showing some novel apps) and afterwards we have a round of 5 X 15 minutes presentations each of other speakers with 5 minutes for discussion, with 15 minutes final round up discussion. We have three of these short presentations from visitors outside Uix:

2) Tatu Harviainen: Tatu Harviainen is currently working as a research scientist at the augmented reality team of VTT Technical Research Centre of Finland. His research work has focused on different aspects of virtual and augmented reality and novel user interfaces. He has M.Sc. on computer science from University of Helsinki and MA on new media from University of Art and Design Helsinki and is currently looking forward in furthering his post graduate studies at Helsinki University of Technology.

3) Mika P. Nieminen from Helsinki University of Technology: http://stratus.soberit.hut.fi/profile/userprofile-5.html

4) Tommi Ilmonen from Multitouch Oy and founding developer of CityWall, Helsinki amongst many other applications.

And two short presentations from projects within Uix:

5) Celine Coutroux and Ivan Avdouevski: Large multitouch public interfaces for artistic installations (CALLAS project) and a jazz festival (S3 project).

6) Toni Laitinen from IPCity (with IPCity team): Worlds of Information: Design and Problems

The running order will be: (times include discussion)
10:00 am - 11:00 am: Hyowon Lee: Designing Novel Applications Inspired by Emerging Media Technologies
11:00 am- 11:20am: Tatu Harviainen: Multitouch development at the VTT's Augmented Reality team
11: 20 am- 11:40am: Mika Nieminen: Holodeck: Controlling context and user experience for design - a research project proposal
11:40am- 12:00pm: Tommi Ilmonen: Interaction Techniques in Ubiquitous Computing: Multitouch Displays
12:00pm- 12:20pm Celine Coutrix and Ivan Avdouevski : Large multitouch public interfaces for artistic installations (CALLAS project) and a jazz festival (S3 project).
12:20- 12:40 pm Toni Laitinen: Worlds of Information: Design and Problems
Discussion 12.40 until finish (1pm)

Coffee and Biscuits will be available. You are all welcome (please invite others) and there will be some interesting work discussed, so please do come!

More enquiries to ann.morrison(a)hiit.fi: Address: Metsänneidonkuja 4, Spektri Building, Pohjois-Tapiola, Espoo

6 Oct Mini-workshop on machine learning and neural representation by the University of Helsinki

Mini-workshop on machine learning and neural representation Tue 6th October 2009, University of Helsinki (Kumpula)

11:30 Aapo Hyvärinen (University of Helsinki)
Introduction

12:00 Jörg Lücke (Goethe-Universität Frankfurt)
Linear, maximal, and occlusive causes for component extraction
(abstract given below)

13:00 Lunch break (on your own)

14:00 Peter Földiák (University of St Andrews)
Semantic representation in neural codes (abstract given below)

15:00 Coffee break

15:30 Michael Gutmann (University of Helsinki)
Learning models of natural images by noise-contrastive estimation
(abstract given below)

16:30 End of workshop

Location: Lecture hall CK112 (basement, next to Unicafe) Exactum Building, Gustaf Hällströmin katu 2b Kumpula Campus, University of Helsinki

Please register for the workshop by reply email (so that I know how much coffee to order).

Welcome!

Aapo Hyvärinen

-------------------------------------------------------------------------

Abstracts of the talks

Jörg Lücke: Linear, maximal, and occlusive causes for component extraction.
Abstract: In the nervous system of humans and animals sensory data are represented as combinations of the data's elementary components. Such representations are considered as `higher-level' because they allow for a more direct read-out of crucial or at least useful information. In machine learning, algorithms such as independent component analysis
(ICA) or non-negative matrix factorization (NMF) have been extensively studied because they allow for the extraction of such elementary components from unlabelled data. They have consequently been highly successful in the analysis of data and have broadly been applied in science and technology. In my talk I will discuss component extraction with discrete latent variables, i.e., I consider algorithms that try to explain the data by combinations of components with discrete factors.
The novel approaches are developed within the framework of probabilistic generative models, and I will start by studying models that combine their components linearly. A novel training method based on approximate EM will be introduced and it is show that the method can efficiently maximize the data likelihood. I will discuss the applicability of the approach to a broader class of models including those with non-linear combinations of components. As one example a model is introduced in which components combine according to a maximum rule (MCA). As another example, I discuss models that use non-linear combinations according to component positions in depth (Occlusive Components Analysis). For all models, results on applications to artificial benchmarks (e.g., bars
tests) and natural data will be shown and evaluated. Finally, I will discuss the implications of the presented results for biological information processing and draw relations to component extraction in neural networks.

Peter Földiák: Semantic representation in neural codes.
Abstract: The question of how activity patterns of neurons represent objects in the world has so far mainly been addressed by asking the question of how the identity of stimuli be decoded from the neural signals. However, an even more interesting question is how the structure of the relationships between items and categories can be represented in a sparse and explicit neural code. The duality between "sets of objects" and "sets of features" have been extensively studied by the field of lattice theory called "Formal Concept Analysis" (FCA).
FCA is proposed as a useful method for analysing a neural code because of this explicit structure. Examples from monkey inferotemporal cortex will be presented. Probabilistic extensions of FCA, and some possible practical computational applications for categorisation, optimal cacheing and semantic memory systems will be discussed.

Michael Gutmann: Learning models of natural images by noise-contrastive estimation.
Abstract: Noise-contrastive estimation is an estimation principle that we have recently developed to learn sophisticated statistical models.
The idea is to train a classifier to discriminate between the observed data and some artificially generated noise. We have shown that this leads to a consistent (convergent) estimator of the parameters. An advantage of the method is that it can be applied to the estimation of unnormalized models (i.e. models where the density function does not integrate to one). Statistical models of natural images are typically
unnormalized: Examples will be shown where we used noise contrastive estimation to learn models of natural images.
 

28 June - CfP: Workshop on Visual Analytics and Knowledge Discovery with ACM SIGKDD

VAKD '09 - Workshop on Visual Analytics and Knowledge Discovery: Integrating Automated Analysis with Interactive Exploration.

A full-day workshop in conjunction with the 15th ACM SIGKDD Conference on Knowledge Discovery and Data Mining in Paris, France, on 28 June 2009.

Submit papers by 20 April 2009.

www.hiit.fi/vakd09 »
 

2 June - Socio-cognitive modeling

The KULTA project, a collaboration between Helsinki University of Technology (TKK), National Consumer Research Center and Helsinki School of Economics, and the computational cognitive systems research group at TKK organize a multidisciplinary research seminar on socio-cognitive modeling.

The seminar is mainly intended for researchers working in disciplines related to to socio-cognitive modeling including but not limited to sociology, cognitive science, and computer science. Also practitioners interested in these questions are welcome.

     Date:  Tuesday, 2 June, 2009, starting at 9:00

     Place: T Building, Helsinki University of Technology (TKK),
            Konemiehentie 2, Espoo

Programme highlights:

   - Abstraction levels in socio-cognitive modeling
   - Modeling virtual world consumers: Case Habbo
   - Analogies between concepts and practices:
     Processes of emergence, learning, diffusion, and disintegration
   - Challenges and research questions in socio-cognitive modeling
   - Thought and hands-on experiments

Please, find the programme and additional information at

     http://www.cis.hut.fi/research/cog/kulta/soccog09.html

Welcome!

Lunch and refreshments are served and therefore we ask you to register by sending e-mail to Tommi Vatanen (first.last at tkk.fi).

28-29 May - The Emerging Media Practices and Environments

http://p2p-fusion.org/emerging/

The media environment is changing from the rigid channels of the past to a social mediaspace, where audiences become also actors, editors and distributors of content, equipped with new software powered social media tools. With new tools and technologies, new practices are born, which in turn inspire and demand new technological and service developments, as well as redesign of regulations, agreements and laws.

The Emerging Media Practices and Environments symposium will discuss this development, focusing especially on the audiovisual media, and aims to identify opportunities, interests, and challenges and develop recommendations for future developments.

The panel sessions include insights from pioneers of new video practices, technology developers, broadcasting companies, audiovisual archives, rights agencies and other important actors in the media environment, and are hosted by Paula Le Dieu, Tommi Laitio, Sanna Marttila and Kari-Hans Kommonen. All participants of the symposium are invited to take part in the discussions and in the development of a wish list of future developments for the audiovisual media environment and its actors. An Open Space session offers participants the opportunity to present their own relevant work and activities to the community.

The discussions take place in the Lume TV studio at the University of Art and Design Helsinki (TAIK) from 9:30 to 17:30 on Thursday, 28 May, followed by a technology day on the morning of Friday, 29 May.

The event is organized by the Arki research group of the Media Lab Helsinki and the P2P-FUSION project.

Welcome! Please RSVP to Sari Ahvenainen at firstname.lastname@taik.fi