ScaDS Logo

COMPETENCE CENTER
FOR SCALABLE DATA SERVICES
AND SOLUTIONS

Aktuelles

ScaDS & BIGGR @ BIG DATA AI-Summit 28. Feb. 2018

Gemeinsam mit Kollegen der Robert Bosch GmbH und der KNIME.com GmbH konnten erste erfolgreiche Projektergebnisse aus dem ScaDS-Projekt BIGGR einem breiterem Publikum präsentiert werden.

Read more...ScaDS & BIGGR @ BIG DATA AI-Summit

ScaDS Dresden/Leipzig at CeBIT 2018

Also this year, ScaDS Dresden/Leipzig will continue to present Big Data related research topics at the IT conference and exhibition CEBIT, this year at the Hannover Exhebition Center from June 11 - 15 in Hannover. On the shared booth together with other universities of the Free State of Saxony, Saxony-Anhalt and the Free State of Thuringia we will be part of the TU Dresden exhibit and inform the public in particular about recent developments in Big Data application areas. ScaDS Dresden/Leipzig presentations will be accompanied with other research topics form the Center of Informations Services and High Performance Computing (ZIH) of the TU Dresden, the 5G Lab and the Cluster of Excellence Center for Advancing Electronics Dresden (cfaed).

In particular, ScaDS Dresden/Leipzig and ZIH will inform the interested public about highlights in its research activities, including amongst others:

  • Big Data and data intensive computing
  • Data analytics and data life cycle management
  • High Performance Computing, Distributed and Cloud Computing
  • Scalable software tools for application support and energy efficient computing

All visitors and interested attendees of the exhibit are very welcome to visit our boot at CEBIT (a detailed location will be published as soon as available).

ScaDS Dresden/Leipzig at Supercomputing 2017

This year International Conference for High Performance Computing, Networking, Storage and Analysis (Supercomputing Conference, SC17) will take place from 12 to 17 November 2017 in Denver (Colorado, USA) and is the biggest and most important get-together in the area of High Performance Computing (HPC). The conference and convention gives an detailed overview about latest achievements in science and technology as well as informs about new trends and innovations. The ZIH as ScaDS Dresden/Leipzig partner will present at the conference in talks, tutorials and discussions its know how in HPC and data intensive computing (Big Data). Furthermore, ScaDS Dresden/Leipzig will present its research highlights at the ZIH conference booth (#1881) alongside other research activities on the fields of performance analysis, energy efficient computing and data analytics. Currently the ZIH plans to extend its existing HPC computing infrastructure and presents its infrastructure concepts (HPC-DA) to bring together concepts from HPC and data analytics. Further research highlights are  online performance analysis and analysis of applications' I/O behavior. At the ZIH booth, visitors can try out the interactive online analysis prototype Vampir Live and experience new I/O analysis capabilities of Vampir, which is a widely used performance investigation tool for parallel applications. In particular, the I/O behavior of applications is becoming a bottleneck on the way to exascale high-performance computers. Novel I/O systems based on non-volatile memory technologies address this bottleneck. Together with the EU Horizon2020 project NEXTGenIO (http://www.nextgenio.eu/), the I/O analysis capabilities of the tools Score-P and Vampir have been significantly enhanced.

More general information can be found at the conference homepage: http://sc17.supercomputing.org/

 

Das Gene Myers Lab des MPI Dresden (CSBD) und das Computer Vision Lab Dresden erhalten den Preis für das beste Forschungspapier der British Machine Vision Conference 2016.


Das Forschungspapier “Mapping Auto-context Decision Forests to Deep ConvNets for Semantic Segmentation” von David Richmond und Kollegen wurde mit dem renommierten Best Science Paper Award der British Machine Vision Conference (BMVC) ausgezeichnet. Die Konferenz fand 2016 in York statt. Die BMVC gehört zu den internationalen Top-Konferenzen im Bereich Computer Vision. Von 365 eingereichten Arbeiten wurde eine Arbeit mit dem Best Science Paper Award bedacht.
Das Forschungspapier entstand in einer Kooperation zwischen dem Gene Myers Lab des MPI Dresden und dem Computer Vision Lab Dresden von Prof. Carsten Rother der TU Dresden. Die Autoren untersuchten den Zusammenhang zwischen den beiden bedeutendsten Techniken für das Problem der semantischen Segmentierung aus dem Bereich des Maschienenlernens. Die Autoren zeigen, dass beide Techniken ineinander überführt werden können, was die Kombination der jeweiligen Vorteile erlaubt. Insbesondere kann die hohe Flexibilität von Neuronalen Netzen mit der hohen Effizienz von Entscheidungsbäumen verbunden werden.

Weiterführende Informationen sowie der Link zur Veröffentlichung finden sich auf der Konferenz-Homepage: http://www.bmva.org/bmvc/2016/papers/paper144/index.html

 

 

Unterkategorien