BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//hpc.bw - ECPv6.17.0//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://www.hsu-hh.de/hpccp
X-WR-CALDESC:Events for hpc.bw
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Europe/Berlin
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20210328T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20211031T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20220327T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20221030T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20230326T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20231029T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20221027T143000
DTEND;TZID=Europe/Berlin:20221027T160000
DTSTAMP:20230203T092400Z
CREATED:20230119T083739Z
LAST-MODIFIED:20230203T092400Z
UID:450-1666881000-1666886400@www.hsu-hh.de
SUMMARY:Seminar Series: Computation & Data
DESCRIPTION:Andreas Fink (HSU): HPC for Solving Combinatorial Optimization Problems in Logistics: Challenges and Examples \n\n\n\nWhile HPC is established in engineering and science\, this is not yet the case for solving NP-hard combinatorial optimization problems in business administration (e.g. in logistics). Existing optimization methods for such problems often have a sequential flow logic and thus must be adapted. We will discuss if and how mixed-integer mathematical optimization methods and solvers are already able to exploit parallel computing power and we will consider the parallelization of (meta-)heuristic methods. \n\n\n\n\n\n\n\nBenedikt Hein (HSU): Distributed Deep Reinforcement Learning: How a hundred years of experience can be gathered in one day \n\n\n\nOver the past decade\, Deep Reinforcement Learning has received considerable attention in Artificial Intelligence research. Successful training in Deep Reinforcement Learning generally requires millions of interactions with a simulated environment. The massive parallelization of simulated training environments recently enabled the training of an outstanding game AI for the computer game DOTA. This talk presents the techniques\, principles and goals of this kind of parallelization.
URL:https://www.hsu-hh.de/hpccp/event/cd2210/
LOCATION:H1\, room 205\, Holstenhofweg 85\, Hamburg\, 22043
CATEGORIES:Seminar Computation & Data
ORGANIZER;CN="hpc.bw":MAILTO:info-hpc-bw@hsu-hh.de
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20220927T100000
DTEND;TZID=Europe/Berlin:20220927T110000
DTSTAMP:20230217T091823Z
CREATED:20220908T085303Z
LAST-MODIFIED:20230217T091823Z
UID:536-1664272800-1664276400@www.hsu-hh.de
SUMMARY:User Meeting
DESCRIPTION:The user meeting is open to every HSUper user to share experiences. \n\n\n\nPlease send an e-mail to info-cbrz@hsu-hh.de if you wish to participate. Depending on the user size an adequate room will be reserved.
URL:https://www.hsu-hh.de/hpccp/event/user-meeting-2/
CATEGORIES:User Meetings
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20220630T150000
DTEND;TZID=Europe/Berlin:20220630T160000
DTSTAMP:20230203T092409Z
CREATED:20230119T082851Z
LAST-MODIFIED:20230203T092409Z
UID:475-1656601200-1656604800@www.hsu-hh.de
SUMMARY:Seminar Series: Computation & Data
DESCRIPTION:Speakers: Thomas Döbbert and Martin Holters \n\n\n\nTitle: Digital sensor-2-cloud campus platform \n\n\n\nAbstract: Wireless networks support highly flexible manufacturing processes towards the digitization of industrial production (automation). 5G is currently being marketed as universal solution for future wireless communication. However\, also other wireless technologies are beneficial and are offering energy efficient and cost-effective solutions\, even with battery-powered smart sensor devices. In this project\, features of 5G are also combined with IO-Link Wireless (IOLW) with respect to robustness and latency to realize highly demanding safety applications in industrial environments.
URL:https://www.hsu-hh.de/hpccp/event/cd2206/
LOCATION:H1\, Hörsaal 3\, Holstenhofweg 85\, Hamburg\, 22043\, Deutschland
CATEGORIES:Seminar Computation & Data
ORGANIZER;CN="hpc.bw":MAILTO:info-hpc-bw@hsu-hh.de
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20220602T150000
DTEND;TZID=Europe/Berlin:20220602T160000
DTSTAMP:20230203T092412Z
CREATED:20230119T080741Z
LAST-MODIFIED:20230203T092412Z
UID:470-1654182000-1654185600@www.hsu-hh.de
SUMMARY:Seminar Series: Computation & Data
DESCRIPTION:Speaker: Derek Groen from Brunel University London. Derek is a lecturer in simulation and modeling\, and his research is interdisciplinary\, focusing primarily on multiscale modelling and high-performance computing for various applications. \n\n\n\nTitle: Multiscale Migration Modelling using Supercomputers: enabling agent-based modelling forecasts in a world of messy\, inaccessible\, biased and incomplete data \n\n\n\nAbstract: In this talk I will present the multiscale migration modelling approach that we have developed in my group during the EU-funded HiDALGO CoE project (hidalgo-project.eu). The approach aims to forecast where persons displaced abroad by conflict may arrive. This information is useful to NGOs and other humanitarian organizations as it can enable them to prepare refugee camps in advance\, and at the right level of capacity. We apply an agent-based modelling algorithm to construct the simulations\, and have coupled them to a range of other models and data sources (e.g. weather data\, food security data and conflict generators) to try and produce accurate forecasts. Our approach has been validated against data from ~8 different conflicts\, and it is being tested in a collaboration with the Save The Children NGO. Additionally\, it is currently one of the main simulation approaches used in the ITFLOWS project (itflows.eu). \n\n\n\nAs part of this talk\, I will also present some highlights on (a) executing our approach in parallel on large scale supercomputers\, (b) how we can use sensitivity analysis to guide our development directions\, (c) how we try to cope with the imperfect data that is rife in these contexts and (d) how we are currently working on many-objective optimisation algorithms to make automatic camp site selection possible.
URL:https://www.hsu-hh.de/hpccp/event/cd2205/
LOCATION:H1\, Hörsaal 3\, Holstenhofweg 85\, Hamburg\, 22043\, Deutschland
CATEGORIES:Seminar Computation & Data
ORGANIZER;CN="hpc.bw":MAILTO:info-hpc-bw@hsu-hh.de
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20220428T150000
DTEND;TZID=Europe/Berlin:20220428T160000
DTSTAMP:20230203T092418Z
CREATED:20230119T075829Z
LAST-MODIFIED:20230203T092418Z
UID:465-1651158000-1651161600@www.hsu-hh.de
SUMMARY:Seminar Series: Computation & Data
DESCRIPTION:“Infrastructure Information Modelling: Workflow for an overall model infrastructure embedded in spatial base data” \n\n\n\nJens Bartnitzek (A+S Consult GmbH) \n\n\n\nInfrastructure Information Modelling is an approach which requires rethinking in the sense of cultural shift towards a systematic and partnership-based approach to major projects. Essentials from mathematics\, computer science and civil engineering are necessary for describing the workflow reasonable. The seminar shows in a real world example the process of creating a 3D overall model of an infrastructure facility starting with federal spatial base data. We take a deeper look on the relevant data interfaces and technical workflows. After the inventory model is constructed\, we connect different data from specialized models into the overall model for interdisciplinary use cases. The overall model is for a collaborative “3D planning” with complete information linkage. We show how those use cases can be realized over multiple objects\, connecting multiple data formats based on results from multiple software products.
URL:https://www.hsu-hh.de/hpccp/event/cd2204/
LOCATION:H1\, Hörsaal 3\, Holstenhofweg 85\, Hamburg\, 22043\, Deutschland
CATEGORIES:Seminar Computation & Data
ORGANIZER;CN="hpc.bw":MAILTO:info-hpc-bw@hsu-hh.de
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Berlin:20220331T150000
DTEND;TZID=Europe/Berlin:20220331T160000
DTSTAMP:20230203T092421Z
CREATED:20230119T074938Z
LAST-MODIFIED:20230203T092421Z
UID:451-1648738800-1648742400@www.hsu-hh.de
SUMMARY:Seminar Series: Computation & Data
DESCRIPTION:15:00-15:30 Yannis Schumann (HSU): Data-driven Inference of Stencils for Discrete Differential Operators \n\n\n\nPartial differential equations (PDEs) are extensively used across scientific disciplines for modeling and describing various processes under consideration. Finite element\, finite difference or similar methods allow to numerically solve them by discretizing the variables under consideration\, e.g. space and time. In this discretized domain\, differential operators can be approximated by matrices with nonzero coefficients for positions in the neighborhood of the considered point –the so called stencil. \n\n\n\nWe consider the inference of stencil weights for differential operators from linear\, one-dimensional and two-dimensional PDEs using comprehensive regression techniques. Starting with the 1D case\, we show that linear regression using an ordinary-least-squares (OLS) approach is able to recover mathematically meaningful stencils given a full-rank matrix of predictor variables. We discuss\, how regularization techniques can allow the inference of the correct stencils even for rank-deficient matrices. We discuss the impact of noise on the data in both predictor and predicted variables for various noise levels and compare different errors-in-variables approaches to mitigate the inherent consequences of noisy predictors. The presented techniques will be extended to the two dimensional case and applied to problems from physics and hydrogeology. \n\n\n\n\n\n\n\n15:30-16:00 Henrik Steude (HSU): It’s more than a “model.train()” — Modern tools and architectures for ML systems \n\n\n\nThe code to train machine learning (ML) models only covers a small part of the entire complexity required by a production-ready ML system. In particular\, model and data versioning\, reproducibility and scalability represent grand challenges. To address these challenges\, various new tools and technologies have been developed under the umbrella term “ML-Ops” over the last years. \n\n\n\nIn our contribution\, we present a selection of technologies\, that have evolved to be popular in the rapidly developing field of ML-Ops. As a concrete example\, we further present a system architecture for a ML platform\, which is used in the dtec.bw project (K)ISS for ML-based analysis of telemetry data of the international space station.
URL:https://www.hsu-hh.de/hpccp/event/cd2203/
LOCATION:H1\, Hörsaal 3\, Holstenhofweg 85\, Hamburg\, 22043\, Deutschland
CATEGORIES:Seminar Computation & Data
ORGANIZER;CN="hpc.bw":MAILTO:info-hpc-bw@hsu-hh.de
END:VEVENT
END:VCALENDAR