Erasmus+ Info Veranstaltung für Studierende an der TF

Technische Fakultät organisiert erneut Siegerehrung des Landeswettbewerbs „Jugend forscht – Schüler experimentieren“

Sonderkolloquium, Jun.-Prof. Dr. Mattias Heinrich, Uni Lübeck / am 07.07.2017

07.07.2017 von 14:15 bis 15:45

Institut für Informatik,Christian-Albrechts-Platz 4, R.715, 24114 Kiel

Titel: Learning Sparse Binary Features for Medical Image Segmentation of the Abdomen

Abstract: In this talk, we explore the capabilities of sparse binary features for medical image segmentation. Due to insufficient contrast and anatomical shape variations local image patches rarely provide sufficient information for accurate segmentation of abdominal structures. Based on our two recent MICCAI papers, we propose to use long-range binary features to robustly capture the image context. Two different classification strategies are subsequently developed. 

First, a very fast approximate nearest neighbour search based on vantage point forests and Hamming distances between feature strings is presented. The classifier can be learned and applied to new data in few seconds. The approach reaches state-of-the-art performance for larger organs on the VISCERAL3 benchmark.

Second, we develop a deep neural network architecture that combines a local CNN path with a new contextual path that encodes the sparse binary features. Following the ideas from Network-in-Network, 1x1 convolutions are employed to learn the best combination of different binary offset locations. We demonstrate experimentally that this restricted feature extraction in the first layer enables to regularise the network with a huge receptive field and leads to short training times of less than 10 minutes. Using only 1 million trainable parameters, the model achieves a accuracy of 64.5% Dice, which is comparable to the best performing, much more complex deep CNN approach for pancreas segmentation.

Finally, the potential use of learned binary features for other tasks in medical image analysis, such as image registration and disease classification will be discussed.

Prof. Carsten Meyer

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (ET&IT), Prof. Elisabetta Chicca, Uni Bielefeld / am 13.11.2017

13.11.2017 von 17:15 bis 18:15

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Learning in silico beyond STDP

Abstract: Synaptic plasticity empowers biological nervous systems with the ability to learn from experience and adjust to environmental changes. Such abilities are a must for artificial autonomous systems and therefore researchers have been devoting significant efforts to the understanding and modelling of plasticity mechanisms. In particular, the field of neuromorphic engineering focuses on the development of full-custom hybrid analog/digital electronic systems for the implementation of models of biological computation and learning in hardware. I will give a short historical overview of the most important plasticity circuits developed following the approach originally proposed by Carver Mead in the late eighties. Afterwards, I will present recent advancements in this field.

zur Person: Elisabetta Chicca hat einen "Laurea"-Abschluss (M.Sc.) in Physik von der Universität Rom 1 ("La Sapienza", Italien 1999) einen Ph.D. in Naturwissenschaften von der Eidgenössischen Technischen Hochschule Zürich (ETHZ, Abteilung Physik) und einen Ph.D. in Neurowissenschaften von dem neurowissenschaftlichen Zentrum Zürich (beides 2006).

 Prof. Chicca forschte als Postdoktorandin (2006-2010) und als Gruppenleiterin (2010-2011) am Institut für Neuroinformatik (Universität Zürich und ETH Zürich) an der Entwicklung von Neuromorphen Signalverarbeitungs- und sensorischen Systemen. Seit 2011 leitet sie die Gruppe Neuromorphic Behaving Systems an der Universität Bielefeld (Faculty of Technology and Cognitive Interaction Technology Center of Excellence, CITEC). Ihre aktuellen Interessen liegen in der Entwicklung von VLSI-Modellen kortikaler Schaltkreise für Gehirn-inspirierte Schaltungen, Lernen in spiking VLSI-neuronalen Netzen und Memristor basierten Systemen, bioinspirierter Wahrnehmung (Olfaktion, aktive Elektrolokalisierung, Hörvermögen) und Motorsteuerung. Die von ihr verwendeten Verfahren zur Herstellung kundenspezifischer Schaltungen (ASICs), sind solche, die auch von unseren neuen Kollegen Andreas Bahr und Robert Rieger genutzt werden.

 In ihrem Vortrag wird Elisabetta Chicca einen kurzen historischen Überblick über die wichtigsten neuromorphen Systeme geben, die nach dem Ansatz von Carver Mead in den späten achtziger Jahren entwickelt wurden. Danach wird sie die neuesten Fortschritte in diesem Bereich vorstellen.

PD Martin Ziegler

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (ET&IT), M.Sc. Jonas Sauter, Nuance Communications / am 20.11.2017

20.11.2017 von 17:00 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Artificial Bandwidth Extension for Speech Signals Using Deep Neural Networks

Abstract: In mobile communication, the bandwidth of transferred speech signals is either narrow-band (300Hz – 3.4kHz) or wide-band (50Hz – 7kHz or higher). As the limitation to 3.4kHz degrades the speech quality and intelligibility, it is of great interest to artificially extend narrow-band speech signals to wide-band speech.

This talk presents a deep neural network (DNN) approach to artificial bandwidth extension with a focus on robustness in practical applications.

It is based on the source-filter model which decomposes the signal into two parts: an excitation signal and a spectral envelope. The excitation (source part) describes the fine spectral structure which consists of white noise for unvoiced speech and an impulse train for voiced speech. The spectral envelope (filter part) describes the coarse spectral structure, i.e. the formants or resonance frequencies that make up different phonemes.

While the extension of the excitation signal can be done with simple mathematical methods that do not introduce strong artifacts, the envelope is much more relevant for the quality of the reconstructed wide-band signal. That is why the wide-band envelope is estimated with DNNs in this approach, which are trained on a large speech corpus.

Short biography

Jonas Sautter studied Electrical Engineering, Information Technology and Computer Engineering at RWTH Aachen University, Germany. He received his Master of Science degree in 2016. The Master’s thesis with the title “Digital Robust Control for Active Noise Cancellation in Headphones and Hearing Aids” was composed at the Institute of Communication Systems at RWTH Aachen. Since November 2016, he is a PhD student at Nuance Communications in Ulm, supervised by Professor Gerhard Schmidt, Head of the Digital Signal Processing and System Theory group at Christian-Albrechts-Universität, Kiel.

Prof. Schmidt

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (ET&IT), Dr.-Ing. Vasudev Kanade Rajan / am 27.11.2017

27.11.2017 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Digital Road Noise Cancellation System Through Active Noise Control

Abstract: The application of active noise cancellation in real-world has not been fully realized yet. From reducing environment noise through the usage of headphones, to engine noise on commercial jets there are a number of use cases. Each of these use case brings its own set of challenges which can be understood only through multi-disciplinary work. One such use case the the reduction of road noise in vehicles. Structure-borne road noise dominates the cabin of modern vehicles. Several road noise cancellation (RNC) prototype systems have been implemented and demonstrated. These systems are based mainly on analog sensors. The placement of these sensors has been so far been based on random optimization methods. In this talk I will talk about the challenges in developing a generic digital RNC system which includes problem analysis, sensor placement, and performance. An adaptive algorithm process the acceleration signals with high convergence and reaction time for various speed and surface ranges, in order to maintain high audible effects for the passengers. Several modern vehicle platforms are integrated with the digital RNC system with ANC microphone at the headliners and the standard audio loudspeaker setup in order to integrate the technology with the existing audio layout of the vehicle.

Short biography

Vasudev Kandade Rajan received Bachelors degree in Electronics and Communication from Visvesvaraya Technological University, Bangalore, India. He joined as Project Research Assistant in July 2008 in the Electrical Communication Engineering Dept, Indian Institute of Science, Bangalore. There he worked on performance management of IEEE 802.11 WLANs until Sept 2009. He then went to obtain his Masters degree (MSc.) in Digital Communications, 2011 and PhD degree in Signal Processing, 2017 from Universtiy of Kiel, Germany. Currently he is working in the R&D department of Harman Becker Automotive Systems GmbH, Straubing, Germany.

Prof. Schmidt

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (ET&IT), Dr. Andreas Bisplinghoff, Cisco / am 04.12.2017

04.12.2017 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: From Long-Haul to Data-Center-Interconnect - Effiziente Signalverarbeitungsalgorithmen für Flexible Optische Netze

Abstract: First coherent optical communication systems, operating at 40Gbps, have been deployed in 2005. Since then, coherent optical technology has undergone remarkable development in the most recent years. Nowadays, state of the art products support line rates up to 400Gbps per wavelength. Next generation systems will primarily target for higher integration density but will presumably also reach line rates of 600Gbps and beyond.

With progress both in CMOS technology and of optical components, coherent optical transmission continuously pushes forward to highest reach for ultra-long haul applications as well as to highest capacity for shorter reach data-center interconnects. Both directions involve very specific requirements both on the capabilities of applied DSP algorithms and integration density of electrical and optical components.

Each new CMOS technology node facilitates the implementation of more sophisticated DSP algorithms. Many DSP components have undergone tremendous development during the most recent years, enabling coherent systems operating at highest transmission rates. Very efficient equalizer algorithms compensate for linear (CD, PMD) and non-linear (SPM, XPM) signal distortions, enhanced soft-decoded forward error correction schemes improve the noise tolerance, and with probabilistic constellation shaping performance will ultimately approach the Shannon limit.

This talk gives a high-level overview about state of the art DSP algorithms and most recent developments in coherent optical communication. It then discusses the balancing act to address the specific requirements of highest reach as well as highest capacity transmission within a single ASIC. Finally, some selected tradeoffs in algorithm and architecture optimization are shown by means of soft-decoded forward error correction as an example.

Bio: Andreas Bisplinghoff was born in Forchheim in 1984. He received the Dipl.-Ing. and Dr.-Ing. degrees both in electrical and electronic engineering from the Friedrich-Alexander University of Erlangen in 2009 and 2015, respectively.

From 2010 to 2013, he was a Research Assistant with the Institute of Microwaves and Photonics at the University of Erlangen. Since 2013 he has been a Hardware Engineer in Advanced Development with the Cisco Optical GmbH. His research interests include the development of slip-reduced carrier phase recovery techniques and of power-efficient forward error correction schemes for coherent optical communication. Andreas Bisplinghoff has broad experience in complexity-aware algorithm design, FPGA-based prototyping, and power-optimized ASIC implementation.

Prof. Pachnicke

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (MaWi), Prof. Friedrich Frischknecht, Universität Heidelberg / am 11.12.2017

11.12.2017 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Malaria transmission: new insights from in vivo imaging and materials science

Abstract: Malaria parasites are transmitted by mosquitoes and multiply to large numbers in red blood cells to cause disease. When taken up by mosquitoes the parasite develop in large cysts in their guts from where they emerge to colonize the salivary glands of the insect. From there the parasites are injected into the skin where they have to migrate to find a blood vessel. After entering the blood vessel the parasites first enter liver cells where they multiply without causing disease symptoms. In this seminar I will illuminate some of this curious biology using a set of different microscopy techniques, molecular genetics and biophysical approaches. We will see how parasites form within the cysts, how they leave the cysts, move within salivary glands and in the skin. Using laser tweezers we have measured the force they can produce and using micro-pillar arrays we have asked whether they adapted their shape in order to find the small blood capillaries into which they enter. For more info see: www.sporozoite.org

Prof. Selhuber

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (MaWi), PD Dr. Pavel Levkin, Karlsruhe Institute of Technology / am 18.12.2017

18.12.2017 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Designing biofunctional interfaces: from superhydrophobicity to cell microarrays

Abstract: Patterns of different surface properties are ubiquitous in nature and serve various important purposes. Desert beetles exploit superhydrophilic spots on their superhydrophobic back to collect water from the morning mist in the desert. Hydrophilic spots on a superhydrophobic surface of lichen plants allow them to uptake water, but also prevent the formation of water layers on the surface that could interfere with the discharge of lichen spores into the air. Superhydrophobic and omniphobic surfaces possess various unique properties including self-cleaning, liquid repellent and cell repellent properties. We are interested in creating precise two-dimensional micropatterns of apparently incompatible and opposite properties such as superhydrophobicity and superhydrophilicity or slippery and adhesive properties. To create such patterns we develop surface coatings with special wettabilities and photochemical surface functionalization strategies. Combining seemingly opposite properties in micropatterns leads to functionalities non-existent on the original homogeneous interfaces. For example, we showed that superhydrophobic-superhydrophilic patterned surfaces could be used to create patterns of cells, arrays of microdroplets suitable for high-throughput cell screenings, formation of arrays of hydrogel micropads or free-standing hydrogel particles with defined shapes for 3D cell culture. Patterned liquid-infused interfaces could be also used to form cell microarrays or arrays of isolated biofilm colonies for biofilm screenings.

Bild

Prof. Selhuber

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (ET&IT) - muss leider entfallen - Prof. Frank Vollmer, University of Exeter / am 08.01.2018

Kolloquiumsvortrag (MaWi), Prof. Weller, Uni Hamburg / am 15.01.2018

Kolloquiumsvortrag (ET&IT), Prof. Martjin van den Heuvel, Brain Center Rudolf Magnu Utrecht / am 22.01.2018

22.01.2018 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Exploring the human connectome

Abstract: Using network science as a general framework to study the network architecture of nervous system connectivity, more and more studies have highlighted the human and animal brain to display features of an efficient communication network. In my talk I will discuss potential general principles of wiring of connectome organization, principles conserved across species, and which may play an important role in general nervous system functioning. I will highlight findings that show connectomes to display cost-effective wiring, pronounced community structure, short communication relays, and the existence of richly connected 'hub regions'. I will discuss theories on how these themes of wiring may play a role in brain disorders, as well as establish a putative link between the micro- and macroscale organization of the human brain in health and disease.

Prof. Kohlstedt

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (MaWi), Prof. Kläui, Uni Mainz / am 29.01.2018

29.01.2018 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Abstract: In our information-everywhere society IT is a major player for energy consumption. Novel spintronic
devices can play a role in the quest for GreenIT if they are stable and can transport and manipulate
spin with low power. Devices have been proposed, where switching by energy-efficient approaches,
such as spin-polarized currents is used [1], for which we develop new highly spin-polarized materials
and characterize the spin transport using THz spectroscopy [2]. Firstly to obtain ultimate stability, topological spin structures that emerge due to the
Dzyaloshinskii-Moriya interaction (DMI) at structurally asymmetric interfaces, such as chiral domain
walls and skyrmions with enhanced topological protection can be used [3-5]. We have investigated in
detail their dynamics and find that it is governed by the topology of their spin structures [3]. By
designing the materials, we can even obtain a skyrmion lattice phase as the ground state of the thin
films [4]. Secondly, for ultimately efficient spin manipulation, we use spin-orbit torques, that can transfer
more than 1ħ per electron by transferring not only spin but also orbital angular momentum. We
combine ultimately stable skyrmions with spin orbit torques into a skyrmion racetrack device [4],
where the real time imaging of the trajectories allows us to quantify the novel skyrmion Hall effect
[5]. Finally to obtain efficient spin transport, we study graphene and low damping ferro- and
antiferromagnetic insulators as spin conduits for long distance spin transport [6] and explore the
superfluid spin current regime in antiferromagnets [7]. We find that we can control magnonic spin
currents by a newly developed magnon spin valve device [8].

[1] Reviews: O. Boulle et al., Mater. Sci. Eng. R 72, 159                                                                                                     
(2011); G. Finocchio et al., J. Phys. D: Appl. Phys. 49,
423001 (2016); A. Bisig et al., PRL 117, 277203 (2016)
[2] M. Jourdan et al., Nature Commun. 5, 3974 (2014);
Z. Jin et al., Nature Phys. 11, 761 (2015).
[3] F. Büttner et al., Nature Phys. 11, 225 (2015).
[4] S. Woo et al, Nature Mater. 15, 501 (2016).
[5] K. Litzius et al., Nature Phys. 13, 170 (2017).
[6] A. Kehlberger et al., Phys. Rev. Lett. 115, 096602 (2015);
S. Geprägs et al., Nature Commun. 7, 10452 (2016).
[7] Y. Tserkovnyak and M. Kläui, arxiv:1707.01082
[8] J. Cramer et al., arxiv:1706.07592

Kläui

Prof. MC Cord

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Kolloquiumsvortrag (MaWi), Frau Dr. Berit Zeller-Plumdorf, Helmholtz Zentrum Geesthacht / am 05.02.2018

Kolloquiumsvortrag (ET&IT), Dr. Timm Faulwasser, Karlsruhe Institute of Technology / am 12.02.2018

12.02.2018 von 17:15 bis 18:00

Institute Ostufer, Geb. D, "Aquarium", Kaiserstr. 2, 24143 Kiel

Titel: Recent Progress on Distributed and Stochastic Optimization for Power Systems

Abstract: The increasing need for the de-carbonization of energy supply calls for new operational methods for power systems. In this context, tailored system and control approaches are pivotal- The specific challenges include the consideration of volatile renewable generation, uncertain forecasts thereof, and highly nonlinear system behavior.

In this talk, we focus on the so-called Optimal Power Flow (OPF) problem, which refers to a class of large-scale non-convex steady-state optimization problems frequently arising in power systems. For example, OPF problems provide optimal set points for power dispatch that satisfy the power flow equations and technical limitations such as generation and/or transmission limits. However, OPF problems are highly non-convex and subject to considerable uncertainties, which includes forecasts of renewable generation and household consumption, line parameters and grid topology.

After a concise problem statement, we provide an overview of the state of the art techniques to considering uncertainties in OPF problems and their bottlenecks. Moreover, we will discuss the concept of Polynomial Chaos Expansions (PCE) which allows to consider non-Gaussian uncertainties in OPF problems. PCE builds upon a series expansion of random-variables. We will present recent results on PCE for convex DC-OPF problems and non-convex AC-OPF problems [1, 2]. Moreover, we will comment on the quantification of PCE truncation errors [3].

Due to their large-scale nature, the distributed solution of OPF problems is subject to considerable research efforts. Thus, we will also comment on our recent results on the distributed solution of OPF problems [4].

References

[1] Mühlpfordt, T.; Faulwasser, T.; Roald, L. & Hagenmeyer, V. Solving optimal power flow with non-Gaussian uncertainties via polynomial chaos expansion. 56th IEEE Conference on Decision and Control, 2017. To appear.

[2] Mühlpfordt, T.; Faulwasser, T. & Hagenmeyer, V. Solving stochastic AC power flow via polynomial chaos expansion. IEEE International Conference on Control Applications, 2016, 70-76.

[3] Mühlpfordt, T.; Findeisen, R.; Hagenmeyer, V. & Faulwasser, T. Comments on Quantifying Truncation Errors for Polynomial Chaos Expansions. arXiv:1708.07655.

[4] Engelmann, A.; Mühlpfordt, T.; Jiang, Y.; Houska, B. & Faulwasser, T. Distributed AC optimal power flow using ALADIN. 20th IFAC World Congress, 2017.

Bio Sketch

Timm Faulwasser has studied Engineering Cybernetics at the University Stuttgart, with majors in systems and control and philosophy, where he graduated 2006. In 2007 he joined the group of Rolf Findeisen at the Institute of Automation Engineering at the Otto-von-Guericke University Magdeburg, Germany. From 2008-2012 he was a member of the International Max Planck Research School for Analysis, Design and Optimization in Chemical and Biochemical Process Engineering Magdeburg. In 2012 he obtained his PhD (with distinction) from Faculty of Electrical Engineering and Information Engineering, Otto-von-Guericke University Magdeburg, Germany. 2013-2016 he was with the Laboratoire d’Automatique, Ecole Polytechnique Fédérale de Lausanne, Switzerland. Since April 2015, he is with the Institute for Applied Informatics at the Karlsruhe Institute for Technology, where he leads the Optimization and Control Group.

His main research interests are optimization-based and predictive control of nonlinear systems with applications in energy systems, mechatronics/robotics, physics, process systems engineering and climate economics.

Prof. Meurer

Diesen Termin meinem iCal-Kalender hinzufügen

zurück

Aktuelle Meldungen

Forschung & Projekte

  • Logo Erneuerbare Energien und Klimaschutz Schleswig-Holstein Biomagnetic Sensing Nina Logo
    Software Challenge Germany Logo
    Kosse SH Logo

     

    FOR2093
    Power Girls and Power Boys Online Mathematik Brückenkurs (OMB+)  Jufo Landeslogo
      Materials for Brain  

     

Kalender

« März 2018 »
Mo Di Mi Do Fr Sa So
26 27 28 1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31 1