Aktuelle Publikationen

März 2021

A Physical Layer for Low Power Optical Wireless Communications

Malte Hinrichs, Volker Jungnickel, Ronald Freund, Jonas Hilt, Anagnostis Paraskevopoulos, Pablo Wilke Berenguer, Dominic Schulz, Peter Hellwig, Kai Lennert Bober

With the goal of enabling optical wireless communications for mobile devices, we assess a physical layer based on high-bandwidth on-off keying modulation. This allows for amplifier designs that avoid operation in a resistive mode, reducing their energy usage to a fraction. Link-level simulations show that the investigated physical layer can deal with typical frontend limitations and can operate in challenging non-line-of-sight channels.


März 2021

Example-Based Facial Animation of Virtual Reality Avatars using Auto-Regressive Neural Networks

Wolfgang Paier, Peter Eisert, Anna Hilsmann

We present a hybrid animation approach that combines example-based and neural animation methods to create a simple, yet powerful animation regime for human faces. We introduce a light-weight auto-regressive network to transform our animation-database into a parametric model. During training, our network learns the dynamics of facial expressions, which enables the replay of annotated sequences from our animation database as well as their seamless concatenation in new order.


März 2021

Radiation pattern of planar optoelectronic antennas for broadband continuous-wave terahertz emission

Simon Nellen, Martin Schell, Björn Globisch, Robert Kohlhaas, Lars Liebermeister, Milan Deumer, Sebastian Lauck, Garrit William Johannes Schwanke

In future wireless communication networks at terahertz frequencies, the directivity and the beam profile of the emitters are highly relevant since no additional beam forming optics can be placed in free-space between the emitter and receiver. We investigated the radiation pattern and the polarization of broadband continuous-wave (cw) terahertz emitters experimentally and by numerical simulations between 100 GHz and 500 GHz. The emitters are indium phosphide (InP) photodiodes with attached planar antenna, mounted on a hyper-hemispherical silicon lens and integrated into a fiber-pigtailed module. As both packaging and material of the emitter was identical for all devices, similarities and differences can be directly linked to the antenna structure. We found that the feeding point structure that connects photodiode and antenna has a large influence on the radiation pattern. By optimizing the feeding point, we could reduce side lobes from ?2 dB to ?13 dB and narrow the 6dB beam angle from ±14° to ±9° at 300 GHz.


März 2021

Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications

Wojciech Samek, Klaus-Robert Müller, Sebastian Lapuschkin, Grégoire Montavon, Christopher J. Anders

This paper provides a timely overview of the field of explainable artificial intelligence (XAI). It explains the theoretical foundations of interpretability algorithms, outlines best practice aspects, demonstrates successful usage of XAI in selected application scenarios, and discusses challenges and possible future directions of this active emerging field.


März 2021

A new concept for spatially resolved coherent detection with vertically illuminated photodetectors targeting ranging applications

Pascal Rustige, Patrick Runge, Martin Schell, Francisco M. Soares, Jan Krause

This paper proposes a novel approach for coherent detection with double-side vertically illuminated photodetectors. Signal and local oscillator are injected collinearly from opposite sides of the photodetector chip. The concept inherently provides angular selective detection and can be used for developing compact, solid-state receiver modules for coherent light detection and ranging (LiDAR).


Februar 2021

A Unifying Review of Deep and Shallow Anomaly Detection

Lukas Ruff, Klaus-Robert Müller, Wojciech Samek, Grégoire Montavon, Jacob R. Kauffmann, Robert A. Vandermeulen, Marius Kloft, Thomas G. Dietterich

This paper gives a comprehensive overview over classic shallow and novel deep approaches to anomaly detection. We identify the common underlying principles and provide an empirical assessment of major existing methods that are enriched by the use of recent explainability techniques. We present specific worked-through examples together with practical advice and discuss open challenges.


Februar 2021

Pruning by Explaining: A Novel Criterion for Deep Neural Network Pruning

Seul-Ki Yeom, Klaus-Robert Müller, Wojciech Samek, Alexander Binder, Sebastian Lapuschkin, Simon Wiedemann, Philipp Seegerer

This paper proposes a novel criterion for CNN pruning inspired by neural network interpretability: The most relevant units, i.e. weights or filters, are automatically found using their relevance scores obtained from concepts of explainable AI (XAI). By exploring this idea, we connect the lines of interpretability and model compression research.


Februar 2021

Optoelectronic frequency-modulated continuous-wave terahertz spectroscopy with 4 THz bandwidth

Lars Liebermeister, Martin Schell, Simon Nellen, Björn Globisch, Robert Kohlhaas, Steffen Breuer, Milan Deumer, Sebastian Lauck

Time-domain spectroscopy with terahertz frequencies typically requires complex and bulky systems. Here, the authors present an optoelectronics-based, frequency-domain terahertz sensing technique which offers competitive measurement performance in a much simpler system.


Januar 2021

InP-Components for 100 Gbaud Optical Data Center Communication

Patrick Runge, Martin Möhrle, Martin Schell, Ute Troppenz, Marko Gruner, Tobias Beckerwerth, Hendrik Boerma

Externally modulated DFB lasers (EML) and vertically illuminated photodetectors are presented. Because of their excellent high-speed behavior and operation wavelength of 1310 nm, the devices are of interest for intra-data center communication. Since the EML and the photodetector chips are compatible with current systems, these devices are candidates for upgrading existing transceivers to higher baud rates. Therefore, a proof of concept for 100 GBaud data transmission with the presented components is demonstrated. Even without predistor! tion, the experiments show clearly open eye diagrams.


Januar 2021

Robustifying Models Against Adversarial Attacks by Langevin Dynamics

Vignesh Srinivasan, Klaus-Robert Müller, Wojciech Samek, Shinichi Nakajima, Arturo Marban

This paper proposes a novel, simple yet effective defense strategy for adversarial attacks on deep learning models. Our algorithm, called MALA for DEfense (MALADE), is applicable to any existing classifier, providing robust defense as well as off-manifold sample detection. In our experiments, MALADE exhibited state-of-the-art performance against various elaborate attacking strategies.



Ergebnisse pro Seite10ǀ20ǀ30
Ergebnisse 1-10 von 45
<< < 1 2 3 4 5 > >>