Die Institute für Informatik, Geoinformatik und Wirtschaftsinformatik arbeiten gemeinsam an der Resilienz von intelligenten Städten (smart cities) und umliegenden Regionen.
Insbesondere aufgrund der stetig steigenden Zahl an Stressfaktoren wie zum Beispiel Klimaveränderungen, extremen Wetterereignissen, wirtschaftlichen Störfaktoren und schnellem Bevölkerungswachstum ist Resilienz eine wesentliche Eigenschaft. Diese soll sicher stellen, dass Dienste und Services aus verschiedenen Bereichen auch unter schwierigen Bedingungen aufrecht erhalten werden können.
Hierzu organisieren die drei Institute ein regelmäßiges Resilienzkolloqium:
2022:
Liebe Kolleg*innen,
im Rahmen unseres Resilienzkolloquiums hören wir am 21.12.2022 um 14:00 Uhr von Prof. Dr. Angela Schwering aus dem Institut für Geoinformatik einen Vortrag zum Thema:
„Resiliency with Physical Computing“
Abstract:
Physical computing, i.e. the implementation of software on hardware interacting with the real world, provides us with many possibilities to implement our own smart systems for resilient applications. In this presentation I will introduce a learning environment that trains computational thinking and computational innovation in the context of real-world environmental problems in the science domain. I will demonstrate how our learning environment supports 21st century skills such as data, computational and scientific literacy, but at the same time address the issue of environmental responsibility of people in lifelong learning.
Zoom-Zugangsdaten: Link
Meeting-ID: 658 2986 9437
Kenncode: 947795
Wir freuen uns wie immer über eine rege Teilnahme.
Viele Grüße
Carina da Silva und Anne Remke
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
Terminänderung:
Vortrag am 09.11.2022 (statt am 26.10.2022) um 14:00 Uhr von Jun.-Prof. Dr. Tanya Braun:
Inference Techniques for Resilience
Abstract:
According to Avizienis et al. (2004), resilience is a system’s ability to remain operational – although at potentially lower operational levels – when exposed to stressors and to adapt its functioning if those stressors persist. When working with formal models in the background, probabilistic inference may be used to predict or identify stressors, or compute necessary adaptations to formal models for a better representation of the system under duress. This leads to an algorithmic technical side to resilience. The inference part should keep going even if there is a sudden influx in observations or queries, or it resume as fast as possible if indeed a change in the model has occurred. This talk looks at resilience for inference. Specifically, it highlights how methods from statistical relational AI can help build more resilient algorithms using its inherent characteristics to support a system's overall resilience.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
Liebe Kolleg*innen,
im Rahmen unseres Resilienzkolloquiums hören wir am 31.08.2022 um 14:00 Uhr von Prof. Dr. Tobias Brandt aus dem Institut für Wirtschaftsinformatik einen Vortrag zum Thema:
„ For cities to become resilient, we need to leave the Smart City behind“
Abstract:
In this talk I will discuss ongoing work related to the opportunities for the Information Systems discipline in strengthening cities’ resilience in face of the growing climate crisis. I will first outline how the smart city label is often used as a marketing vehicle, selling IT solutions without tackling the real challenges cities are already facing. Secondly, I will outline outcome-oriented research opportunities for the IS discipline that focus on using digital means to strengthen cities’ resilience. I will particularly emphasize three focal areas, namely environmental resilience (outcomes related to the physical impact of climate change such as heat and flooding), social resilience (outcomes related to the social impact of climate change such as refugee migration and social tension), and administrative resilience (outcomes related to strengthening public administrations’ capabilities to manage a rapidly transforming world).
Zoom-Zugangsdaten: https://wwu.zoom.us/j/67218967068?pwd=eVNoVk9CR1M4bkVpR1dMVzg2T1Ztdz09
Meeting-ID: 672 1896 7068
Kenncode: 881773
Wir freuen uns wie immer über eine rege Teilnahme.
Viele Grüße,
Carina da Silva
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
Liebe Kolleg*innen,
unser nächster Termin des Resilienzkolloquiums findet am 15.06.2022 um 14:00 Uhr statt. Wir hören einen Vortrag von unseres Kollegen Julius Adelt zum Thema:
„ Towards Safety and Resilience in Intelligent Hybrid Systems with Deductive Verification “
Abstract:
Deductive verification is a powerful approach for establishing crucial safety properties of intelligent hybrid systems. However, deductive verification requires abstract formal descriptions, e.g. a formal model, properties, contracts, and invariants, which are often not available in industrial design processes and hard to obtain for the unpredictable, trial and error learning processes of reinforcement learning. In addition, resilience often involves dynamic properties such as recovery and the safe adaptation to external stressors. In this talk, we summarize our work on deductive verification of safety and resilience of intelligent hybrid systems that are modeled with industrial design languages like Simulink.
Zoom-Zugangsdaten: https://wwu.zoom.us/j/66541280209?pwd=c2Q3bzhmalBrR3VmOWM5dGM1RXZPUT09
Meeting-ID: 665 4128 0209
Kenncode: 688067
Wir freuen uns wie immer über eine rege Teilnahme.
Viele Grüße,
Carina da Silva
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
Liebe Kolleg*innen,
im Rahmen unseres Resilienzkolloquiums hören wir am 11.05.2022 um 14:00 Uhr von Prof. Dr. Mariëlle Stoelinga (Radboud University Nijmegen /University of Twente) einen Vortrag zum Thema:
„Predictive Maintenance via Fault Tree Analysis and Model Checking“
Abstract:
Predictive maintenance is a promising technique that aims at predicting failures more accurately, so that just-in-time maintenance can be performed, doing maintenance exactly when and where needed. Thus, predictive maintenance promises higher availability, fewer failures at lower costs.
In this talk, I will advocate a combination of model-driven (esp. fault trees) and data analytical techniques to get more insight in the costs versus the system performance (in terms of availabillity, reliability, remaining useful lifetime) of maintenance strategies. I will show the results of three case studies from railroad engineering namely rail track (with Arcadis), the HVAC (heating, ventilation, airco; with Dutch railroads).
I will also go into recent developments on learning fault trees and rare event simulation.
Bio:
Prof. Dr. Mariëlle Stoelinga is a professor of risk management, both at the Radboud University Nijmegen, and the University of Twente, in the Netherlands.
Stoelinga is the project coordinator on PrimaVera, a large collaborative project on Predictive Maintenance in the Dutch National Science Agenda NWA. She also received a prestigious ERC consolidator grant. Stoelinga is the scientific programme leader of Risk Management Master, a part-time MSc programme for professionals. She holds an MSc and a PhD degree from Radboud University Nijmegen, and has spent several years as a post-doc at the University of California at Santa Cruz, USA.
Zoom-Zugangsdaten: https://wwu.zoom.us/j/66737857874?pwd=OEVPRzdRcitmR1duSllXWG9XMmxtdz09
Meeting-ID: 667 3785 7874
Kenncode: 107731
Wir freuen uns über eine rege Teilnahme.
Viele Grüße,
Carina da Silva
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resilienzkolloquium - Save the Date
Terminänderung
Liebe Kolleg:innen,
leider müssen wir den angekündigte Vortrag des Resilienzkolloquiums von Jana Seep vom 16.02.2022 auf den 02.03.2022 um 14:00 Uhr verschieben. Die Zoom-Zugangsdaten bleiben dieselben (siehe unten).
Wir hören einen Vortrag von unserer Kollegin Jana Seep zum Thema:
„Modellierung und Analyse semantisch annotierter Trajektorien “
Zoom-Zugangsdaten: https://wwu.zoom.us/j/67599077198?pwd=aS9MN1lIUTF1eXh1b3pxR2UxeGl0dz09
Meeting-ID: 675 9907 7198
Kenncode: 173141
Wie immer freuen wir uns über eine zahlreiche Teilnahme.
Viele Grüße,
Carina
Abstract:
Wenn man die Bewegung einer Menge von Objekten analysieren möchte, betrachten herkömmliche Verfahren häufig nur die geometrischen Muster der Bewegungen. Um aber die Bewegung eines Objektes wirklich zu verstehen, müssen wir auch die verschiedenen externen und internen Faktoren berücksichtigen, die diese beeinflusst haben könnten. Diese Informationen sind in sogenannten semantischen Trajektorien enthalten, allerdings wird die Analyse dieser Daten bislang nur sehr elementar durch Algorithmen unterstützt.
Semantische Trajektorien bestehen aus Punkten, die, neben der Information über den geometrischen Ort, mit beliebig vielen Attributen jeder Art, die die Umgebung, aber auch die Eigenschaften des Objektes beschreiben können, annotiert sind. Solche Trajektorien werden zum Beispiel bei der Beobachtung von Tieren, Menschen oder bestimmten Wetterphänomenen aufgezeichnet. Durch die Analyse der semantischen Trajektorien erhoffen sich Domänenexperten tiefere Einblicke in komplexe Bewegungsmuster oder auch die Möglichkeit neue Zusammenhänge zwischen Trajektorien zu erkennen.
In diesem Vortrag stellen wir unseren Ansatz für die Analyse semantischer Trajektorien vor. Mit diesem ist es uns möglich, die Einflüsse verschiedener Parameter auf die Bewegung zu erkennen und die gegebenen Trajektorien anhand ihrer semantischen Unterschiede zu klassifizieren. Dazu berechnen wir ein Modell einer Menge von Trajektorien, definieren ein semantisches Ähnlichkeitmaß und stellen Algorithmen vor, die das Modell nutzen, um Trajektorien zu clustern oder eine semantisch repräsentative Trajektorie eines Clusters zu berechnen. Wir stellen abschließend Anknüpfungspunkte für den Einsatz unserer Verfahren zur Verarbeitung annotierter Daten in den Bereichen “Mobility” oder “Economy” vor.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
19.01.2022 um 14:00 Uhr
Marina Evers und Prof. Dr. Lars Linsen (Institut für Informatik)
“Interactive Visual Data Analysis in the Context of Resilience”
Abstract: In a resilient system, monitoring is required in order to detect and react to stressors and eventually to adapt. Monitoring requires data collection and analysis. A user-centric data analysis allows the analysts to react based on their domain expertise. Interactive visual analysis methods provide means towards such a user-centric approach. In this talk, we will present an overview and examples for interactive visual data analysis approaches in the context of resilient hinterland, focusing on the domains of mobility and environment. We will exemplify the concept of interactive visual data analysis for environmental data by presenting in more depth an approach for uncertainty-aware visualization of regional time series correlation.
2021:
Liebe alle,
ich würde euch gerne auf einen Vortrag am kommenden Dienstag aufmerksam machen, welcher im Rahmen des ERCIS Lunchtime Seminars (https://www.ercis.org/research/ercis-lunchtime-seminar) stattfinden wird. Die Details findet ihr unten. Cosmin und ich arbeiten seit mehreren Jahren zusammen, unter anderem im Bereich HPC und Satellitendaten. Cosmin ist auch einer der Entwickler von Futhark (https://futhark-lang.org/).
Viele Grüße und eine schöne Restwoche!
Fabian
---
When: Tuesday, 14.12.2021, 12:15-13:00 (Münster time; UTC+1, Central);
waiting room opens at 12:00
Where: Virtual Meeting
Zoom Link:
https://wwu.zoom.us/j/67477067440?pwd=enFmelNtTGRpNDQ0ZjlRb3oxR01GUT09
Meeting ID: 674 7706 7440
Passcode: 703922
Title: Reverse AD for an Array Language with Nested Parallelism -
Redundant Execution: A Practical Backpropagator
Abstract: Automatic differentiation (AD) is a practical way for
computing derivatives of functions that are expressed as programs. AD
has been recognized as one of the key pillars of the current machine
learning (ML) revolution, and has key applications in other domains such
as financial algorithms, computational fluid dynamics, atmospheric
sciences, and engineering design optimization. This talk presents a
technique for applying reverse AD on a higher-order functional array
language, which allows programs to be written as a nested composition of
parallel operators such as map, reduce, scan (also together with loops
with in-place updates). We start by implementing the "tape" (required by
reverse AD) by means of a code-generation technique that performs
redundant computation, but which still preserves the work-depth
asymptotic of the original program. More importantly, our technique
revels a very useful time-space tradeoff that can be navigated by
traditional compiler transformations: Essentially, strip-mining a look k
times increases the re-execution overhead by a factor of k but
exponentially decreases the memory footprint of checkpointing. Having
designed the glue that binds scopes together, we then discuss how to
efficiently differentiate standard parallel constructs such as map,
scan, (multi-)reduce. We report the implementation of both the forward-
and reverse-mode AD as code transformations in the optimising compiler
of the Futhark language, which allows to generate high-performance code
for parallel execution on GPUs. We present an experimental evaluation of
our technique on nine benchmarks, including the applications from
AD-Bench, but also a recurrent neural network (LSTM), and k-means. We
compare against other state-of-the-art implementations, each tested on
their own home ground, and demonstrate competitive performance on both
sequential CPU and parallel GPU execution.
Bio (http://hjemmesider.diku.dk/~zgh600/): Cosmin Oancea is an Associate
Professor with the Department of Computer Science (DIKU), University of
Copenhagen. His main research interests lie in the general field of
programming language design and implementation, with focus on compiler
optimizations aimed at parallel execution on highly-parallel hardware.
He is one of the main architects of the optimizing compiler of the
Futhark functional language. Previous to that, he has studied with Alan
Mycroft (University of Cambridge) and Lawrence Rauchwerger (Texas A & M)
various compiler analyses---ranging from static to entirely
dynamic---related to automatic parallelization of (imperative)
loop-based code. Cosmin holds a PhD from the University of Western
Ontario, where he has studied topics related to language
interoperability under the supervision of Stephen M. Watt.
24.11.2021, 14:00 - 15:30 Uhr via Zoom (Link folgt)
Decentralised Energy Management: Challenges and Opportunities
Prof. Johann Hurink, Utwente, Niederlande
https://people.utwente.nl/j.l.hurink
15.06.2021 um 12:00 Uhr
Taylor Faith, Department of Geography, King’s College London
Messy Maps: exploring flood risk and resilience in the urban Global South using qualitative GIS
Abstract: This seminar will present results from an interdisciplinary project looking at how we put a more inclusive definition of resilience on the map. Towns and cities across Sub Saharan Africa are rapidly urbanising, and now 1 in 10 of the world's population live in slums or informal settlements. These slums are increasingly exposed and vulnerable to hazards such as flooding and fire, yet are often 'missing from the map'. In this project, we use a range of qualitative GIS techniques such as immersive 360 degree Storyspheres and spatial social network maps to visualise what resilience means to low-income neighbourhoods in Nairobi (Kenya) and Cape Town (South Africa). I will also reflect on my experience as a quantitative physical geographer coming to grips with qualitative methods and the challenges/opportunities this has brought, especially in the era of Covid-19 disruptions to data collection!
24.03.2021 um 13:00 Uhr
Prof. Xiaoyi Jiang (Institut für Informatik)
Graph (Network) Analysis: A Pattern Recognition Perspective
2020:
17.12.2020 um 14:15 Uhr
Prof. Dr. Christian Kray (Geoinformatik)
“Resilience and Pedestrian Navigation Support”
24.11.2020 um 12:15 Uhr
Herr Adam Widera (Wirtschaftsinformatik)
“Practitioner-centered Assessment of Socio-technical Innovations in Crisis Management: Insights from the DRIVER+ Demonstration Project”
14.10.2020 um 14:00 Uhr
Prof. Dr. Anne Remke (Informatik)
„Dependability, resilience and security in local energy distribution”