Sprachen

L3S Research Center -  RSS Feeds

Sie sind hier

Startseite » Veranstaltungen » Ausgewählte Themen der Informatik

Informatik Kolloquium: Aktuelle Themen der Informatik

Prof. Dr. Steffen Staab

Semantics Reloaded

Steffen Staab, Uni Koblenz

In phase 1 of semantic technologies, research focused on explicitly modeling the semantics of data and knowledge in ontologies and likewise models. In phase 2 of semantic technologies, people tried to learn semantics of data from how it was used with other data by means of terminology inducation and ontology learning using machine classification and clustering. While both of these approaches remain valid undertakings, we posit that one should now also consider a phase 3 of semantic technologies. In this phase, semantics arises from what we do with data and knowledge.

We give some, very heterogeneous, examples from our current projects:

  1. Studying users in social networks with transfer learning
  2. Studying the semantics of user interactions with eye tracking
  3. Studying the usage of data in programs with logical inference.

10 November 2017, 14:00
Multimedia Room, 15th Floor, L3S

Prof. Dr. Bettina Berendt

Why formalizing fairness won't fix algorithmic discrimination

Bettina Berendt, KU Leuven

‘‘Big Data’’ and data-mined inferences are affecting more and more of our lives, and concerns about their possible discriminatory effects are growing. Methods for discrimination-aware data mining and fairness-aware data mining aim at keeping decision processes supported by information technology free from unjust grounds. However, these formal approaches alone are not sufficient to solve the problem. In the present article, I describe reasons why discrimination with data can and typically does arise through the combined effects of human and machine-based reasoning, and argue that this requires a deeper understanding of the human side of decision-making with data mining. I describe results from a large-scale human-subjects experiment that investigated such decision-making, analyzing the reasoning that participants reported during their task to assess whether a loan request should or would be granted. I derive data protection by design strategies for making decision- making discrimination-aware in an accountable way, grounding these requirements in the accountability principle of the European Union General Data Protection Regulation, and outline how their implementations can integrate algorithmic, behavioral, and user interface factors.

24 November 2017, 14:00
Multimedia Room, 15th Floor, L3S

Dr. Bastian Pfleging

Automotive User Interfaces in the 21st Century: Challenges to Support Drivers in the Transition from Manual to Automated Driving

Bastian Pfleging, Ludwig-Maximilians-Universität München

Our daily life becomes more and more digital through the integration of digital, connected, and ubiquitous systems. Even the car is much more than a vehicle to get from A to B: Besides conducting the primary task of driving the car, drivers want to perform a multitude of non-driving-related activities (e.g., operate the infotainment system, communicate with the outside world etc.) at the same time. The imminent risk is that such activities distract the driver and, thus, put drivers, passengers, and the environment at risk. Therefore, one important challenge is the exploration and design of automotive user interfaces which enable safe, diverse, exciting, and easy-to use means to perform also additional activities in the car.
With the transition to automated driving, non-driving-related activities become even more important and will be one major selling point and field of innovation for future cars. One important aspect will be the driver-vehicle interface which provides enough freedom to perform a multitude of activities (e.g., office work, communication, relaxation, media consumption) but still ensures a safe fallback to manual driving. In my talk, I will present our research activities in this domain of automotive user interfaces and provide insights into our current projects. This includes interaction aspects, driver monitoring, concepts for non-driving-related activities, and aspects of automated driving.

8 December 2017, 17:00
Multimedia Room, 15th Floor, L3S

Prof. Dr. Volker Markl

Mosaics in Big Data: Stratosphere, Flink and beyond

Volker Markl, TU Berlin

The global database research community has greatly impacted the functionality and performance of data storage and processing systems along the dimensions that define “big data”, i.e., volume, velocity, variety, and veracity. Locally, over the past five years, we have also been working on varying fronts. Among our contributions are: (1) establishing a vision for a database-inspired big data analytics system, which unifies the best of database and distributed systems technologies, and augments it with concepts drawn from compilers (e.g., iterations) and data stream processing, as well as (2) forming a community of researchers and institutions to create the Stratosphere platform to realize our vision. One major result from these activities was Apache Flink, an open-source big data analytics platform and its thriving global community of developers and production users. Although much progress has been made, when looking at the overall big data stack, a major challenge for database research community still remains. That is, how to maintain the ease-of-use despite the increasing heterogeneity and complexity of data analytics, involving specialized engines for various aspects of an end-to-end data analytics pipeline, including, among others, graph-based, linear algebra-based, and relational-based algorithms, and the underlying, increasingly heterogeneous hardware and computing infrastructure. At TU Berlin, DFKI, and the Berlin Big Data Center (BBDC), we aim to advance research in this field via the Mosaics project. Our goal is to remedy some of the heterogeneity challenges that hamper developer productivity and limit the use of data science technologies to just the privileged few, who are coveted experts.

19 January 2018, 14:00
Multimedia Room, 15th Floor, L3S

Prof. Dr.-Ing. habil. Wolfgang Schröder-Preikschat

Predictability Issues in Operating Systems

Wolfgang Schröder-Preikschat, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)

Predictability is always subject to the underlying assumptions being made. For real-time systems, time response of processes in relation to the strictness of deadlines is of particular importance. With an additional focus on embedded systems, space and energy requirements become relevant as well and need to be considered in combination. As far as software is concerned, structure and organisation of the programs to be executed determines whether or not predictable processes will take place in a given computing system. Design for predictability is an overarching aspect that crosscuts the whole computing system and particularly addresses operating systems.

This talk is about structuring principles of non-sequential programs - in the shape of but not limited to operating systems - to abet predetermination of quality attributes of non-sequential (real-time) processes, it is not about analytical methods to effectively predetermine these attributes. Issues in operating systems as to space, timing, and energy requirement are touched. Emphasis thereby is on coordination of cooperation and competition between processes, namely synchronisation. It is shown how measures of process synchronisation against the background of many-core processors cater to these issues.

29 Juni 2018, 15:00
Multimedia Room, 15th Floor, L3S

Prof. Dr. Klaus-Robert Müller (Photo:  FhG FIRST)

Machine Learning and AI for the Sciences - Towards Understanding

Prof. Dr. Klaus-Robert Müller, Technische Universität Berlin

In recent years, machine learning (ML) and artificial intelligence (AI) methods have begun to play a more and more enabling role in the sciences and in industry. In particular, the advent of large and/or complex data corpora has given rise to new technological challenges and possibilities. In his talk, Müller will touch upon the topic of ML applications in the sciences, in particular in neuroscience, medicine and physics. He will also discuss possibilities for extracting information from machine learning models to further our understanding by explaining nonlinear ML models. E.g. Machine Learning Models for Quantum Chemistry can, by applying interpretable ML, contribute to furthering chemical understanding. Finally, Müller will briefly outline perspectives and limitations.

4 Mai 2018, 14:00
Multimedia Room, 15th Floor, L3S

Dr. Esfandiar Mohammadi

Anonymity Trilemma: Anonymity, Low Latency, and Low Bandwidth Overhead - Choose Two

Dr. Esfandiar Mohammadi, ETH Zürich

Anonymous communication networks (ACNs) constitute a basic building block for privacy-preserving protocols in the Internet by protecting connection meta-data, such as the IP address of the sender or the receiver of a message. Many ACNs proposed in the literature incur a high communication cost, either with a high bandwidth overhead or with a high latency compared to communication without the ACN. This talk considers a strong notion of anonymity where ACNs have to provide anonymity under unlimited usage of the ACN against strong network-level observers. Our main result is that there are two cost dimensions for ACNs for which there exist a joint cost constraint: latency (leading to a latency overhead) and a dummy message rate (leading to a bandwidth overhead). The cost constraint states that for ACNs that aim to achieve strong anonymity the latency overhead can only be low if the dummy message rate is high and, vice versa, the dummy message rate can only be low if the latency overhead is high.

21 März 2018, 14:00
Multimedia Room, 15th Floor, L3S

Prof. Dr. Hermann de Meer

ELECTRIFIC - Towards Grid-Friendly EV Charging

Prof. Dr. Hermann de Meer, University of Passau

Electric mobility leads to an increasing challenge for power gird operators, particularly due to its irregular and unknown load profile. Power grid enhancements are considered either as cost-intensive or as environmentally unfriendly and, hence, more intelligent ICT-based solutions are needed for economic and ecological reasons. Therefore, our intention is to develop a practical approach of grid-friendly smart Electric Vehicle (EV) charging methods. The approach entails methods, namely: (i) Proactive EV charging control via prediction of available charging capacity and a corresponding intelligent scheduling of charging processes; Availability of renewable energy sources as well as (EV) load volatility control can be incorporated. (ii) Reactive, decentralized charging process control as a response to critical grid situations. Proactive forecasting of available power capacity and energy from (distributed) renewable sources can lead to a better utilization of the power grid in place and extend the usage of renewable energy, which is required for a successful transition in energy policies. A reactive control of ongoing charging processes guarantee that the power grid infrastructure can run at its limits, while increasing the power quality in the grid. This dual concept exploits the flexible potential of the power supply network and at the same time optimizes the ongoing charging processes to meet the requirements of the grid.

6 Juli 2018, 15:00
Multimedia Room, 15th Floor, L3S

Dr.-Ing. Axel Wendt

Perception of autonomous systems

Dr.-Ing. Axel Wendt (Robert Bosch GmbH)

The talk will be about perception systems of autonomous systems, with focus on a project of Bosch and Daimler working together to take the development of fully automated, driverless vehicles forward (http://www.bosch-presse.de/pressportal/de/en/bosch-and-daimler-are-worki...). The objective is the joint development of software and algorithms for an autonomous driving system. The project will take the comprehensive vehicle expertise of Daimler – the world’s leading premium-class automaker – and combine it with the systems and hardware expertise of Bosch, the world’s biggest automotive supplier. The synergies that arise as a result will be channeled into making this technology ready for production as early as possible. In promoting a system for fully automated, driverless vehicles for city driving, Bosch and Daimler want to improve urban traffic flows, enhance road safety, and provide an important building block for the traffic of the future. Among other things, the technology will increase the attractiveness of car sharing. It will allow people to make optimum use of the time they spend in cars, and open up new ways of being mobile to people without driver’s licenses. The idea behind it is that the vehicle will come to the driver completely autonomously in cities, not the other way around.

10 Juli 2018, 16:00

A104, Nienburger Str. 1, Hannover

(Eckeingang gegenüber der Haltestelle „Schneiderberg“ der Stadtbahnlinien 4 und 5)

Dr. Minas Dasygenis

A survey on current research work on OpenCL

Dr. Minas Dasygenis,University of Western Macedonia

Every day more and more data are being created, processed and archived raising the need for efficient computational efficiency. The only way to tackle this gargantuan amount of data is through parallelization and efficient usage of the current multithreaded architectures. There are various frameworks or computing paradigms to help a researcher towards this goal. Amongst all these, the state of the art is the OpenCL computational framework. In this talk, we will investigate the maturity and current status of OpenCL, together with research results on real-world problems. We will discuss different scenarios of OpenCL, either with a local or cloud-based job scheduler OpenCL accelerator, various accelerated systems in data-intensive domains like computer vision, weather simulation, neural networks, proposed OpenCL design flows, rendering, image manipulation, automatic source-to-source OpenCL compilers, high level OpenCL synthesis, and the unification of GPU and CPU and DSP and FPGA platforms as processing cores. In the end, we will draw conclusions about the future of OpenCL and acceleration in general, and how can somebody benefit from this programming model.

15 Juni 2018, 14:00
Multimedia Room, 15th Floor, L3S