direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

Subject Area Experts

The founding members of the Data Analytics Laboratory at TU Berlin include Volker Markl, Klaus-Robert Müller, Odej Kao, Anja Feldmann, Thomas Wiegand, Thomas Sikora, and Jean-Pierre Seifert.


Areas of Expertise / Active Research Areas
Volker Markl
The prior work of Volker Markl lies in the area of database management systems, architectures, and query processing algorithms. This includes in particular work on multidimensional indexing, applying machine learning algorithms to query optimization, statistics management, declarative specification, compilation, optimization, and parallelization of iterative data flow programs, and big data benchmarking. Volker is leading the Stratosphere project, a collaborative research unit by the German National Science Foundation (DFG) that focuses on building a big data analytics infrastructure, as well as several European and national research efforts related to Stratosphere. In the context of Stratosphere Volker focuses on data analysis programming languages, the automatic parallelization and optimization of data analysis algorithms in the areas of machine learning, algebraic optimization and text mining, as well as big data benchmarking. He has further conducted work in information integration (the DAMIA project) and is currently conducting work in new hardware architectures for information management (the SindPad and RETHINK projects), as well as information marketplaces (the MIA and DOPA projects).
Klaus-Robert Müller
Previous work of Klaus-Robert Müller encompasses machine learning and a number of applications. He has worked on kernel methods, neural networks, large scale learning, signal processing, and in application areas like the brain computer interface, computer vision, and bioinformatics. He has studied methods for the reliable detection of outliers in complex event streams, which relates to the veracity aspect in this program. Computer vision and signal processing will be important for the different media types encountered in this project. Finally, large scale and kernel learning provide important prior work to tackle questions of large scale learning.
Odej Kao
Previous work of Odej Kao includes massive-parallel resource management for highly efficient processing of semi-structured data and data streams on high-performance and cloud platforms. The major focus is set on cross-optimized booking and assignment of resources for efficient query and stream processing. Already developed systems such as OpenCCS (for massively-parallel compute clusters) or Nephele (for cloud-based platforms) are used for many years in production environments and serve as staring points for large research projects such as HPC4U, AssessGrid, Stratosphere and many others. During these projects additional features such as advanced methods for fault tolerance, failure prediction, job migration, and virtualization were developed and integrated. Additional related work considers the platforms for plug-and-play of medical devices and online analysis of the delivered data streams. The results of the analysis are transferred to hospitals for online assistance. Analogous projects are performed in the energy of smart grids and integration of renewable energies. Odej Kao is a member of the Stratosphere Collaborative Research Unit. In the research unit, Odej has researched and developed a platform for large­-scale data analysis in infrastructure-as-a-service settings from the aspects of fault-tolerance and processing data analysis programs in these massively-parallel settings.
Anja Feldmann
Working towards an improved understanding of the Internet has been the research focus of Anja Feldmann over the past 15 years. For example, this includes work on understanding the Internet Ecosystem, current traffic streams in the Internet, the nature of Internet traffic, tools for network intrusion detection, and also work on the current architectural limitations of the Internet and how to overcome them relying on information from the network, and/or using software defined networking principles, and/or relying on network virtualization.
Thomas Sikora
Description/analysis as well as compression of graphics, images, audio, and video has been prime research focus of Thomas Sikora over the last twenty years. This includes significant work on novel concepts for compression of audio and 2D/3D video data and on analysis/interpretation of audio, images, and video for multimedia retrieval and surveillance applications. As the video chairman of the ISO MPEG group he was responsible for the development of the MPEG-4 video compression standard as well as for the MPEG-7 visual standard targeted for visual content description. Especially, the MPEG-7 standard can be seen as instrumental for sparking the enormous research momentum worldwide on multimedia content description, search, filtering, and classification.
Thomas Wiegand
Prior work of Thomas Wiegand has addressed video coding, multimedia networking as well as semantic approaches to media analysis and coding. In the area of video coding, Wiegand made significant contributions to the video coding standards H.264/MPEG4-AVC and H.265/MPEG-HEVC, extended the compression to scalable formats and to stereoscopic as well as autostereoscopic 3D video. Moreover, together with Klaus-Robert Müller, he investigated novel methods for measuring subjective perception of visual differences using EEG. His work on multimedia networking, includes novel work on error-resilient transmission of real-time video, contributions to standardization in IETF for RTP payloads, specifications for video broadcast to televisions and handsets in DVB, and extensions to MPEG-DASH . He has conducted pioneering research in the field of semantic approaches to video coding.
Sebastian Möller
The Quality and Usability Lab headed by Sebastian Möller is interested in understanding human behavior and judgment when interacting with technology. The goal is to develop methods, algorithms and tools for improving human-machine interaction, and human-human-interaction mediated by technology. For this purpose, data is collected either in the laboratory or in the field, and put into relation to user judgments on quality and usability. For data collection in the field, mobile apps have proven to be a powerful tool, and the lab has successfully developed apps for administering student life (Mobile Campus Charlottenburg, MoCCha), organizing conferences or events (Interspeech and DAGA conference apps, Long Night of Sciences app), or supporting a user’s private life (energy-saving app Sense4En). Data collected with these apps is currently analyzed for understanding user behavior, and offering new services of added value. In addition, Sebastian Möller develops algorithms for predicting speech and audio-visual quality, and supports the international ITU-T standardization work in this area.
Klaus Obermayer
We are concerned with the principles underlying information processing in biological systems. On the one hand we want to understand how the brain computes, on the other hand we want to utilize the strategies employed by biological systems for machine learning applications. Our research interests cover three thematic areas. Models of neuronal systems: In close collaboration with neurobiologists and clinicians we study how the brain processes visual information and perception is linked to cognitive function. Research topics include: cortical dynamics, representation of visual information, adaptation & plasticity, reward-based learning and decision making in uncertain environments and how its interaction with perception and memory. Machine Learning and Neural Networks: Here we investigate how machines can learn from examples in order to predict and act. Current research topics include the learning of representations & semi-supervised learning schemes, prototype-based methods, and models of decision making & reinforcement learning. Analysis of neural data: Here we apply statistical methods and machine learning to the analysis of multivariate biomedical data. Research topics vary and currently include the analysis of multimodal data, for example, correlating behavioral, neuroanatomical, brain imaging, and genetic data.


Your application for an individual "consultation hour" with one of our "Data Science and Engineering" advisors, please send/give to:   

Claudia Gantzer, +49 314 23555 / room EN-728 / email: 

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions

Data Science and Engineering Track Contact

Juan Soto
+49 30 314 23551
Office: E-N 724