INESC-ID

About: INESC-ID is a nonprofit organization based out in Lisbon, Portugal. It is known for research contribution in the topics: Computer science & Context (language use). The organization has 932 authors who have published 2618 publications receiving 37658 citations.

Emergent Stories Facilitated

Abstract: In this paper we address the issue of how to transform a set of actions and behaviors generated by autonomous characters into a meaningful and interesting story. To do that, we consider the issues involved in developing an architecture that is generic enough to be applied to entertainment/educational applications that convey stories through the interactions between intelligent graphical characters and a user. As such, issues in organizing the emergent content are discussed and a generic story facilitator (a kind of game master) is described. Finally two storytelling applications that use the architecture are described demonstrating the generality of the approach.

A 64-channel ASIC for TOFPET applications

Abstract: A 64-channel ASIC for TOF PET imaging is presented The circuit provides time and energy measurements of events produced by a SiPM coupled to a L(Y)SO fast scintillator This ASIC is developed in the framework of the EndoTOFPETUS collaboration as an option for the readout of external 200×200 mm plate detector, which consists of 3×3×15 mm crystals and 3×3 mm (active area) SiPMs Using the chip with non-segmented and/or higher light yield crystals is possible The same applies for photodetectors with different gain, polarity, or even higher dark count rate The targeted 200 ps timing resolution for the system and the need for a low power consumption have driven the choice of a closed-loop amplifier input stage and a 50 ps time binning TDC based on analogue interpolation A power consumption between 5 to 10 mW per channel is expected to guarantee a SNR of at least 20 dB for the single photon, using a SiPM with 320 pF terminal capacitance

Efficient development of data migration transformations

Abstract: In this paper, we present a data migration tool named DATA FUSION. Its main features are: A domain specific language designed to conveniently model complex data transformations; an integrated development environment that assists users on managing complex data transformation projects and an auditing facility that provides relevant information to project managers and external auditors.

Constructing virtual 3D models with physical building blocks

Abstract: Constructing virtual 3D models typically requires specialized desktop modeling tools (e.g., CAD tools), which, while very powerful, tend to require a lot of precision, time, and expertise from the user. We present StereoBlocks, a system that combines a Kinect depth camera with 3D stereoscopic projector to allow the user to build complex virtual 3D models from available physical objects. By treating the camera information as a continuous 3D digitizer, we are able to capture the details of the real world and re-project virtual objects side-by-side to real objects. The user is able to visualize such mixed reality model through stereoscopic projected imagery tightly aligned with the real world. In our system, it is literally possible to build the entire virtual castle, using only a single physical brick piece. We discuss our prototype implementation and report on early feedback from the four users that evaluated our system.

Propagation of Uncertainty Through Multilayer Perceptrons for Robust Automatic Speech Recognition.

Abstract: Observation uncertainty techniques offer a way to dynamically compensate automatic speech recognizers to account for the information missing in real world scenarios. These techniques have been demonstrated to effectively be able to compensate multiple environment distortions and improve the integration of ASR systems with speech enhancement pre-processing through uncertainty propagation. Unfortunately observation uncertainty techniques rely on statistical methods and as such are limited to GMM-HMM architectures. In this paper we explore the application of observation uncertainty and uncertainty propagation techniques to multi-layer perceptrons (MLPs). We develop solutions for propagation through a generic MLP and exemplify potential gains with an large vocabulary robust ASR experiment on the AURORA4 database using an Hybrid MLP-HMM recognizer.