Startseite > Symposium2022 > Keynote Speakers

Keynote Speakers

Explainable AI

Keynote Speaker: Prof. Dr. Wojciech Samek (Fraunhofer Heinrich Hertz Institute Berlin)

more information

From Attribution Maps to Concept-Level Explainable AI

The emerging field of Explainable AI (XAI) aims to bring transparency to today's powerful but opaque deep learning models. While local XAI methods explain individual predictions in form of attribution maps, thereby identifying “where” important features occur (but not providing information about what they represent), global explanation techniques visualize “what” concepts a model has generally learned to encode. Both types of methods thus only provide partial insights and leave the burden of interpreting the model's reasoning to the user. Building on Layer-wise Relevance Propagation (LRP), one of the most popular local XAI techniques, this talk will connect lines of local and global XAI research by introducing Concept Relevance Propagation (CRP), a next-generation XAI technique which explains individual predictions in terms of localized and human-understandable concepts. Other than the related state-of-the-art, CRP answers both the “where” and “what” question, thereby providing deep insights into the model’s reasoning process. In the talk we will demonstrate on multiple datasets, model architectures and application domains, that CRP-based analyses allow one to (1) gain insights into the representation and composition of concepts in the model as well as quantitatively investigate their role in prediction, (2) identify and counteract Clever Hans filters focusing on spurious correlations in the data, and (3) analyze whole concept subspaces and their contributions to fine-grained decision making. By lifting XAI to the concept level, CRP opens up a new way to analyze, debug and interact with ML models, which is of particular interest in safety-critical applications and the sciences.

Wojciech Samek is a Professor for “Machine Learning and Communications” at the Technical University of Berlin, Head of the Department of Artificial Intelligence at Fraunhofer Heinrich Hertz Institute and Fellow at BIFOLD - Berlin Institute for the Foundation of Learning and Data. He studied computer science in Berlin and Edinburgh, was a visiting researcher at the NASA Ames Research Center, Mountain View, USA, and received the Ph.D. degree with distinction from TU Berlin in 2014. He then founded the Machine Learning Group at Fraunhofer HHI, which he headed until 2020. Dr. Samek is associated faculty at the ELLIS Unit Berlin and the DFG Graduate School BIOQIC, and member of the scientific advisory board of IDEAS NCBR. Furthermore, he is a senior editor of IEEE TNNLS, an editorial board member of PLoS ONE and Pattern Recognition, and an elected member of the IEEE MLSP Technical Committee. He is recipient of multiple best paper awards, including the 2020 Pattern Recognition Best Paper Award, and part of the expert group developing the ISO/IEC MPEG-17 NNR standard. He is the leading editor of the Springer book "Explainable AI: Interpreting, Explaining and Visualizing Deep Learning" (2019), co-editor of the open access Springer book “xxAI – Beyond explainable AI” (2022), and organizer of various special sessions, workshops and tutorials on topics such as explainable AI, neural network compression, and federated learning. Dr. Samek has co-authored more than 150 peer-reviewed journal and conference papers; some of them listed by Thomson Reuters as "Highly Cited Papers" (i.e., top 1%) in the field of Engineering.


Secure and Privacy-Preserving Services

Keynote Speaker: Prof. Dr. Anja Lehmann (Hasso Plattner Institute Potsdam)

more information

Privacy-preserving Authentication for Connected Services

Classic signatures and certificates provide cryptographically strong authentication, yet enforce unique identifiability of the data’s origin. This can pose privacy challenges, in particular when data from devices of end users is collected. Privacy-preserving authentication allows users or devices to authenticate data in a pseudonymous und unlinkable manner, avoiding unique traces. While full unlinkability is preferable from a privacy perspective, the correlation among different data points is often crucial for data analytics. Thus, solutions are needed that guarantee strong privacy and unlinkability as default option when data is collected, yet still allow to selectively link and correlate the minimally required information in a controlled way later on. In this talk, I will give an overview of the cryptographic approaches for such authentication schemes, their applications and open challenges.

Anja Lehmann is a professor at the Hasso-Plattner-Institute, University of Potsdam where she leads the Cybersecurity and Identity Management group. Anja works on the development and analysis of cryptographic protocols with provable security guarantees. Before joining HPI in 2020, she was a researcher in the Cryptography & Privacy group at IBM Research – Zurich and holds a PhD in computer science from Darmstadt University of Technology, obtained in 2010.


Platforms and Markets

Keynote Speaker: Prof. Dr. Jan Krämer (University of Passau)

more information

Economic Regulation of Digital Markets: Prospects and Pitfalls

Digital Markets are characterized by the presence of strong network effects, economies of scale, data-drivenness and a global reach, all of which make them - unlike traditional offline markets - prone to ‚market tipping‘. In a tipped market, a single firm dominates and can exercise market power, which raises numerous concerns with respect to economic efficiency, market contestability and fairness, such as a lower rate of innovation, a lower degree of competition, higher prices or lower consumers’ privacy. To remedy this, policy makers in the EU and elsewhere have proposed a specific regulation for digital markets outlining general obligations and prohibitions for digital gatekeepers. The keynote will review some of the new provisions targeted at digital gatekeepers set out in the EU’s Digital Markets Act, which will come into force in 2023.  Through a mainly economic viewpoint, but also employing a technological and legal perspective, the talk will highlight what the intended goal of the provisions is, but also why some may be ineffective or even counterproductive in practice. The talk will conclude by highlighting that interdisciplinary research on digital markets is both relevant and needed as ever, as many issues remain unsolved and the impact of the new regulatory framework will fundamentally change the market environment.

Jan Krämer is a Professor for Information Systems at the University of Passau, Germany, where he holds the Chair for Internet & Telecommunications Business. He is also the director of the Research Training Group 2720 on “Digital Platform Ecosystems” at the University of Passau funded by the German Research Foundation, and Academic Co-Director at the Centre of Regulation in Europe (CERRE), a Brussels-based academic think tank. He has a diploma degree in Business Engineering and Management, and a Ph.D. in Economics, both from the Karlsruhe Institute of Technology, but he also spent longer research visits and study periods at the Massachusetts Institute of Technology and Boston University. His research interests include the economic regulation of telecommunications and internet markets, as well as digital ecosystems and data-driven business models. On these issues he frequently serves as independent academic expert for governmental institutions, such as the European Commission, the German Ministry of Economic Affairs or the Body of the European Regulators of Electronic Communications (BEREC). He has published numerous books and articles in leading academic journals in the areas of Information Systems and Economics and serves as associate editor for several of these journals.


The Future of Computing

Keynote Speaker: Prof. Dr. Robert Wille (TU München, SCCH GmbH)

more information

Software for Quantum Computing: How to Design and Develop Applications for the Next Big Thing?

Quantum computers are becoming a reality and promise to solve certain tasks that are infeasible to solve on conventional computing machines. Accordingly, both academia but also industry started to heavily invest into this new technology. At the same time, "programming" a quantum computer follows completely different paradigms than those typically used in Computer Science. Moreover, design tasks which are considered trivial for conventional computers suddenly become extremely complex in quantum computing. In this talk, we are discussing how we can address these challenges by utilizing dedicated design automation methods and software tools for quantum computing. To this end, we are covering typical design tasks such as simulation, compilation, and verification of corresponding quantum algorithms and circuits. For more details, please see https://www.cda.cit.tum.de/research/quantum/.

Robert Wille is a Full and Distinguished Professor at the Technical University of Munich, Germany, and Chief Scientific Officer at the Software Competence Center Hagenberg, Austria. He received the Diploma and Dr.-Ing. degrees in Computer Science from the University of Bremen, Germany, in 2006 and 2009, respectively. Since then, he worked at the University of Bremen, the German Research Center for Artificial Intelligence (DFKI), the University of Applied Science of Bremen, the University of Potsdam, and the Technical University Dresden. From 2015 until 2022, he was Full Professor at the Johannes Kepler University Linz, Austria, until he moved to Munich. His research interests are in the design of circuits and systems for both conventional and emerging technologies. In these areas, he published more than 400 papers and served in editorial boards as well as program committees of numerous journals/conferences such as TCAD, ASP-DAC, DAC, DATE, and ICCAD. For his research, he was awarded, e.g., with Best Paper Awards, e.g., at TCAD and ICCAD, an ERC Consolidator Grant, a Distinguished and a Lighthouse Professor appointment, a Google Research Award, and more.


Dynamic, Heterogeneous and Hybrid Networks

Keynote Speaker: Prof. Dr. Anja Feldmann (Max-Planck-Institute Saarbrücken)

more information

Internet Traffic Analysis at Scale

In this talk I will use multiple Internet measurement studies as examples to outline the benefits of performing Internet scale traffic analysis, including Implications of the COVID-19 Pandemic on Internet traffic, Collaborative Detection and Mitigation of Amplification DDoS Attacks, detecting IoT devices through the lens of an ISP, as well as P4-enabled Network-assisted Congestion Feedback.

Anja Feldmann got her Ph.D. from Carnegie Mellon University in 1995. The next four years she did research work at AT&T Labs Research, before taking professor positions at Saarland University, the TU Munich, and TU Berlin. From 2012 to 2018 she served on Supervisory Board of SAP SE. Since the beginning of 2018, Anja is a director at the Max Planck Institute for Informatics in Saarbrücken, Germany. She is a member of multiple academies and on the steering committees of CoNEXT and IMC. She was TPC-chair of Sigcomm, IMC, CoNEXT, as well as HotNets.