Time filter

Source Type

Grachten M.,Austrian Research Institute for Artificial Intelligence OFAI | Krebs F.,Johannes Kepler University
IEEE Transactions on Multimedia

The study of musical expression is an ongoing and increasingly data-intensive endeavor, in which machine learning techniques can play an important role. The purpose of this paper is to evaluate the utility of unsupervised feature learning in the context of modeling expressive dynamics, in particular note intensities of performed music. We use a note centric representation of musical contexts, which avoids shortcomings of existing musical representations. With that representation, we perform experiments in which learned features are used to predict note intensities. The experiments are done using a data set comprising professional performances of Chopin's complete piano repertoire. For feature learning we use Restricted Boltzmann machines, and contrast this with features learned using matrix decomposition methods. We evaluate the results both quantitatively and qualitatively, identifying salient learned features, and discussing their musical relevance. © 1999-2012 IEEE. Source

Payr S.,Austrian Research Institute for Artificial Intelligence OFAI
Applied Artificial Intelligence

The special issue of Applied Artificial Intelligence brings together results of and reflections on the research work undertaken in the project Social Engagement with Robots and Agents (SERA) during the years 2009 and 2010. The description of the setup and the field study is provided by the article, Describing the Interactive Domestic Robot Setup for the SERA Project. The article Theory of Companions: What Can Theoretical Models Contribute to Applications and Understanding of Human-Robot Interaction? begins, as did the project, with a review of social psychological theories of human-human relationships. The article, From Data to Design, addresses reflections on methodology in analyzing human-robot interaction from the engineers' perspective. It takes a critical look at qualitative methods of data analysis and proposes the narrative method as a path from episodic data to design. The final article, On the Nature of Engineering Social Artificial Companions, carries on the development perspective. Source

Jancsary J.,Austrian Research Institute for Artificial Intelligence OFAI | Matz G.,Vienna University of Technology
Journal of Machine Learning Research

We investigate minimization of tree-reweighted free energies for the purpose of obtaining approximate marginal probabilities and upper bounds on the partition function of cyclic graphical models. The solvers we present for this problem work by directly tightening tree-reweighted upper bounds. As a result, they are particularly efficient for tree-reweighted energies arising from a small number of spanning trees. While this assumption may seem restrictive at first, we show how small sets of trees can be constructed in a principled manner. An appealing property of our algorithms, which results from the problem decomposition, is that they are embarrassingly parallel. In contrast to the original message passing algorithm introduced for this problem, we obtain global convergence guarantees. Copyright 2011 by the authors. Source

Payr S.,Austrian Research Institute for Artificial Intelligence OFAI
Proceedings - IEEE International Workshop on Robot and Human Interactive Communication

A field study with a simple robotic companion is being undertaken in three iterations in the framework of a EU FP7 research project. The interest of this study lies in its design: the robotic interface setup is installed in the subjects' homes and video data are collected during ten days. This gives the rare opportunity to study the development of human-robot relationships over time, and the integration of companion technologies into everyday life. This paper outlines the qualitative inductive approach to data analysis, and discusses selected results. The focus here is on the interactional mechanisms of bringing conversations to an end. The paper distinguishes between "closing" as the conversational mechanism for doing this, and "closure" as the social norm that motivates it. We argue that this distinction is relevant for interaction designers insofar as they have to be aware of the compelling social norms that are invoked by a companion's conversational behaviour. © 2010 IEEE. Source

Rank S.,Austrian Research Institute for Artificial Intelligence OFAI
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

In this chapter, a scenario-based analysis of the guiding vision of a virtual butler is presented. After introducing the concept of scenario-based analysis for comparing agent-based technology design, we use the characterization of the scenario hinted at in the vision document to discuss several technological issues that arise from it. By disregarding non-technical issues, we arrive at problems (or rather challenges) of technology in a wide sense that could be steps in the direction of the virtual butler. The order of presentation of these challenges is based on a subjective estimation of the complexity involved in arriving at the competence required for a virtual butler. © Springer-Verlag Berlin Heidelberg 2013. Source

Discover hidden collaborations