Blog

  • till

    October 27, 2020 by

    SDI-C_grafik

    Einladung zur virtuellen Projektvorstellung am 09.11.2020 ab 9:30

    Dear SDIL partners,

    The first micro-projects of the Smart Data Innovation Challenge have been completed and exciting results are available.

    We actually wanted to take the opportunity to organize a networking event in Karlsruhe. Unfortunately, the current situation here has put a damper on our plans, but we’re taking the opportunity to present the results online and meet up with everyone virtually.

    The event with presentations, discussions, technical parts and networking opportunities will take place next Monday on 09.11.2020, 9:30 to 10:00 virtually in gather.town:

    https://gather.town/app/jj2P1N0fsV9zGwhY/SDIC_MEETUP_2020

    Take the opportunity to exchange ideas digitally but still personally. We would be happy to welcome interested participants in any case. Do not hesitate to come by even for a short time (e.g. for a networking session).

    9:00 Admission

    9:30 – 10:00 Welcome

    • “Why Smart Data Innovation Challenges are Important for Research”, Michael Beigl, Speaker of SDIL, KIT
    • “From Big Data to AI from the perspective of a European Public Private Partnership”, Laure Le Bars, VP BDVA / SAP

    10:00-10:45 The role of AI in mobility beyond autonomous driving

    Presentation and discussion of the results from the microprojects

    10:45-11:15 Networking with interactive sessions in gather.town

    • Room for individual discussions
    • Information on the next tenders
    • Posters of funded and unfunded micro-projects
    • Presentation of the current technical platform:
      • Simple use of the HTC cluster under Python with Dask, Till Riedel
      • Quick start to use the GPU cluster, Yexu Zhou
    • Virtual tour data center
    • Introduction https://euhubs4data.eu/
    • … write to office@sdil.de if you want to present here

    11:15-12:00 AI to improve processes: Man vs. Machine?

    Presentation and discussion of the results from the micro projects

    12:00-12:30 Further development of the SDIL platform

    • Cloud-/AI-Services extension for the SDIL HANA platform, Frank Müller, SAP Business Transformation Services
    • Current and planned developments on the SDIL platform, NN , KIT/SCC
    • Graduation, Michael Beigl, KIT

    until about 14:00 further online networking opportunities via gather.town

    You can participate in the event without registration: However, we ask you to indicate your name and in brackets your organization when “entering”. In case of technical problems please contact us at office@sdil.de. (You are welcome to drop by now, I myself will be in the lobby on Tuesday and Thursday at 15:00 each day, if there are any questions about the event, the SDIL in general or about the technology).


  • Ravivanpong

    Aufruf zur Smart Data Innovation Challenge (SDI-C)

    August 26, 2019 by

    SDI-C_logo

    The SDIL sets for a new structure and processing of its work, starting October 2019.

    From October on the BmBF funded research project SDI-C will be a possibility for six AI-micro projects to be able to work in corporation with the SDIL on the SDIL platform. The micro projects will be chosen competitive within all research institutes of KIT, Fraunhofer Gesellschaft, DFKI and Forschungszentrum Jülich. The aim is to develop an idea to its implementation in collaboration with the SDIL, in a short amount of time.

    Streaming analytics are an important part of AI, if fast decisions need to be made or data needs to be continuously evaluated. Funded will be four projects with a reference to high-frequency data streams and time series who use AI-algorithms and -architectures based on industrial relevant data. Furthermore, two thematically open AI-micro projects with pilot character will be funded, given a research impulse with high application potential.

    The projects will have a duration of six months and are desired to involve small and medium-sized companies as associated partners and data providers who bring their own questions and benefit from the results.

    The new AI microprojects are designed to challenge the opportunities of AI based on the use of new industrial data sources in terms of their innovation potential.

    Learn more at: http://sdil.de/en/sdi-c/


  • andreas_wierse

    SDSC-BW with LGI and Logistics at the Big Data.AI Summit in Berlin

    June 10, 2019 by Andreas Wierse

    SDSC_BW_BigDataAiSummit2019

    SDSC-BW was represented together with the potential analysis partner LGI on the Big Data.AI Summit in Berlin 2019.

    This year Bitkom’s Big Data.AI Summit was merged with the hub-berlin event. This does not only led to a rather cool location in Berlin, but also to a significantly higher number of participants (numbers between 5,000 and 10,000 participants were mentioned). The event program was also significantly more extensive, with a keynote by Timotheus Höttges, the CEO of Telekom, which is greatly superior. Other prominent figures included Katharina Barley, Frank Mastiaux, Anja Karliczek and, as a special highlight, Cedric Vilani, Fields medalist and AI advisor to President Macron.

    SDSC-BW was represented at the event together with our potential analysis partner LGI: by Dorothea Küttner from LGI and Andreas Wierse from SICOS BW. As opposed to our last year’s presentation slot at the end of the event, in a large hall with only a handful of people, an unfavourable position, we did have a much better placement on the agenda this time: right in the middle of the day, directly after the lunch break, but at the heart of the entire lecture halls, we were able to attract around 50 listeners, a good number of participants compared to the other parallel presentations.

    Thematically, of course, everything was about logistics, the title was “The Prediction of Logistics Resource Needs Based on Smart Data Analytics”. We shared the presentation, Andreas Wierse presented the Smart Data Solution Center Baden-Württemberg and its principle of operation. Dorothea Küttner then explained in terms of content, what was implemented in the potential analysis and what results we came to. Specifically, it was about the possibility of improving personnel planning, i.e. to be able to better estimate how much staff is likely to be needed a day or even a week in advance. For this purpose, the corresponding data records of orders from nine years were used, about 16 million data sets with 32 features. The KIT experts applied different algorithms to this data, in the end Ada Boost worked best. Especially for the weekly forecast, this approach performed much better than the previous experience-driven approach.

    The audience followed the talk with great interest and afterwardsactually several interested listeners, especially from the logistics sector, came to Mrs. Küttner to learn more about the background (with Mrs. Küttner carefully taking care not to divulge any business relevant information). In summary, this presentation was very successful. The event could be used both for further networking, as well as to see how the topic develops in general. In the accompanying exhibition, some companies and research institutions were represented (the Smart Data Forum was also present), so that you could get a good overview.

    We have already noted the date for the next year: on 1 and 2 April 2020, we want to be present again with an interesting potential analysis from the SDSC-BW.

    Find out more at https://sdsc-bw.de/lgi


  • plamen-kiradjiev

    February 04, 2019 by Plamen Kiradjiev

    IBM_Machine_FrontRack_AC922

    In the future SDIL will have a new IBM platform extension: the IBM Power System Accelerated Computer Server (AC922).

    Designed to meet the challenges of Artificial Intelligence (AI), the AC922 with the latest Power9 processor technology provides the power needed to handle today’s high-performance computing (HPC), analysis, and AI. In particular, for corporate AI, the AC922 enables data-intensive workload. For example, deep-learning frameworks and accelerated databases through reduced and optimized AI training times for reclaiming existing data storage.

    The framework for deep learning and machine learning is made possible by the AI-platform PowerAI Enterprise. The platform provides the suitable framework for rapid AI performance and gaining more precise insights from the accelerated IBM Power Servers.

    The AC922 is the only server capable of providing the process speed of data transaction between CPUs and GPUs (I / O performance) to this extent. With this ability to fully leverage its GPU accelerators by eliminating I / O performance bottlenecks and sharing memory across GPUs and CPUs, the system enables data-intensive workloads for HPC, deep learning, and AI.

    Several institutes of KIT are already interested and intend to use the new workload capacities of artificial intelligence, made possible by the AC922. Elaborate text analysis, multilingual speech recognition and direct textual language translation of audio signals are just a few examples of the work of the TECO and ISL, who also benefits from the possibilities of the new technology.

    Learn more at https://www.ibm.com/en-us/marketplace/power-systems-ac922/details


  • Ravivanpong

    “Digitalisierung Läuft!” visits Karlsruhe

    January 28, 2019 by Ployplearn Ravivanpong

    Digitalisierung_Läuft!

    The event “Digitalisierung Läuft!“ took place with its digitization marathon in IT-strong city of Karlsruhe

    The third stage of the event “Digitalisierung Läuft!” organised by the German minister of the Interior and digitization minister of Baden-Württemberg Thomas Strobl has led to the AI-strong research facility KIT and FZI Karlsruhe. The running group around Thomas Strobl and the ex-runner-up in 100-kilometer Marathon World Championship, Jürgen Mennel, reached one of the most outstanding cities for IT and digitization – Karlsruhe.

    On this third stage one of the topics of discussion has been IT-security as sensible data in case of fitness and health applications plays an essential role. Data scientists of the KIT research group “TECO” equipped some of the runners with sensor technology during the digitization marathon. Diverse live data of the participants such as the respiratory rate and environment information, e.g. temperature, elevation and particulate matter level, were evaluated immediately during the run. After the participants arrived at the FZI research facility of computing science, innovation partner of the Karlsruhe Institute of Technologies (KIT), scientists of the KIT and FZI presented some of their research on artificial intelligence.

    More on (in German) https://www.kit.edu/kit/pi2018146_digitalisierung-lauft-am-ursprung-der-informatik.php