Details Colloquium Spring 2022

24.02.2022 – Insights on Machine Learning for mHealth Data Sources

Speaker: Prof. Dr. Rüdiger Pryss

Host: Prof. Dr. Nathalie Giroud

Abstract

Digital phenotyping, experience sampling, digital health, and ecological momentary assessments (EMA) are only a few methods and strategies that have been recently presented at the intersection of mobile technology, healthcare, medicine, and neuroscience. If these strategies are implemented with powerful technical frameworks, large-scale data sources become possible at rather low collection costs. In the context of chronic diseases, the use of mobile technology in combination with the aforementioned strategies can particularly help to empower patients in coping with their individual health situation more properly. The evaluation of the created data sources, however, increasingly requires special attention. For example, the applied data collection procedure or how users behave when using mobile technology in everyday life have shown to pose new challenges for the data evaluation. One promising direction in this context constitutes the use of machine learning algorithms.
In this talk, based on existing evaluations, pitfalls and possibilities are shown when using machine learning algorithms on data sources that have been generated by the use of mobile technology in the medical domain.

Bio

Rüdiger Pryss studied at the Universities of Passau, Karlsruhe, and Ulm. He holds a Diploma in Computer Science. After graduating, he worked as a consultant and developer in a software company. In 2015, he received a PhD in Computer Science. In his doctoral thesis, Rüdiger focused on fundamental issues related to mobile process and task support. Rüdiger Pryss is experienced with teaching courses on database management, programming, service-oriented computing, business process and document management, mobile application engineering, medical informatics, and data science. Since November 2019, Rüdiger Pryss is full professor of medical informatics at the University of Würzburg. His research is based on next generation mobile information systems and infrastructures to support the medical domain in conducting clinical studies and performing patient interventions. In this context, he also focuses on the application and development of data science methods that can be applied to medical data. On top of this, he investigates methods from cognitive psychology and how they can be utilized to learn more about the cognitive procedures when using complex (mobile) information systems.

 

17.03.2022 – Database-Oriented Approaches for Retrieving Tens of Billions of High-Dimensional Features

Speaker: Prof. Dr. Björn Þór Jónsson

Host: Dr. Luca Rossetto / Prof. Abraham Bernstein, PhD

Abstract

Scalable retrieval of high-dimensional feature vectors is an important component of many applications in multimedia and other fields, but also a very challenging problem. In this talk, we discuss the challenges of high-dimensional indexing at scale, and then present two database-inspired approximate indexing methods designed for large-scale retrieval. We present results from experiments with the two largest feature collections reported in the literature, 28.5 billion SIFT features on a single server and 42.9 billion SIFT features in a distributed setting. Finally, we discuss an application of the latter approach to interactive retrieval over the 99.2 million images of the YFCC100M collection.

Bio

Björn Þór Jónsson is an Associate Professor at the IT University of Copenhagen, Denmark. Björn works in the broad field of Multimedia Analytics, applying multi-dimensional analysis concepts and techniques to large-scale multimedia collections. Previously, Björn studied scalability of multimedia retrieval, where he was involved with the two largest feature vector collections reported in the literature. Björn has a special interest in promoting demonstrations, live events, and reproducibility, e.g. serving as Reproducibility Chair for ACM Multimedia 2019-2020 and ICMR 2021-2022. He served as general co-chair for MMM 2017, CBMI 2019, ACM ICMR 2020 and SISAP 2020, and is currently planning MMM 2022.

 

05.05.2022 – Process Science. The Interdisciplinary Study of Continuous Change

Speaker: Prof. Dr. Jan vom Brocke

Host: Prof. Dr. Gerhard Schwabe / Prof. Abraham Bernstein, Ph.D.

Abstract

It all happens through processes, and it happens in processes: Organizations work through processes, our lives unfold in processes; processes make our economy, and they also shape our environment and society. Process science is an emerging field of science that intends to advance our understanding of processes through the integration of contributions from various academic disciplines, such as computer science, management science, organization science, engineering science or cognitive science. Two reasons constitute the establishment of Process Science: First, we live in times of rapid and continuous change, where processes - not systems – are becoming the predominant phenomenon of interest, and these processes span across diverse (disciplinary) boundaries. Processes constitute a distinct object of research accumulating contributions, and it is important to better understand processes in order to better support both economy and society in dealing with change. Second, digital trace data, combined with advanced data analytics capabilities, offer new and unprecedented opportunities to study processes through multiple data sources. Process science builds on event log data, as collected by sensor technology, enterprise systems or social media to capture and discover real-world processes (descriptive process science); it analyses such data in a rich empirical context to better understand processes, e.g. how economic behavior effects environmental change (explanatory process science), and it also develops innovative solutions to be used in practice to influence change for the benefit of economy and society (prescriptive process science). This talk will introduce into Process Science. It will discuss how process science - by integrating data from different sources and analysing them from different disciplinary perspectives - can significantly improve our understanding of processes and thus make an important contribution to helping industry and society manage change.

Bio

Jan vom Brocke is the Hilti Endowed Chair of Business Process Management and Director of the Institute of Information Systems at the University of Liechtenstein. Jan`s work has been published in many of the A+ and Financial Times Top 50 ranked journals, among others, in Management Science, MIS Quarterly, Journal of Management Information Systems, Journal of Information Technology, Journal of the Association for Information Systems, European Journal of Information Systems, Information Systems Journal, Communications of the ACM, and MIT Sloan Management Review. He is also an Associated Researcher at the National University of Ireland in Galway and an Associated Lecturer at the University of Lucerne and the University of St.Gallen. He has served in many senior academic roles, including as Vice President Research of the University of Liechtenstein, Vice President Education of the Association for Information Systems (AIS) and Vice President Technology of the Association for Business Research (VHB). In recognition of his work, he has been named a Fellow of the AIS and appointed a member of the AIS college of senior scholar.

 

12.05.2022 – On the Reliability of Coverage-based Fuzzer Benchmarking

Speaker: Dr. Marcel Böhme

Host: Prof. Dr. Alberto Bacchelli

Abstract

Given a program where none of our fuzzers finds any bugs, how do we know which fuzzer is better? In practice, we often look to code coverage as a proxy measure of fuzzer effectiveness and proclaim that the fuzzer which achieves more coverage is better.
Indeed, in one of the largest studies to date, we find that a fuzzer that covers more code also finds more bugs. There is a very strong correlation between the coverage achieved and the number of bugs found by a fuzzer. Hence, it might seem reasonable to compare fuzzers in terms of the coverage achieved, and from that derive empirical claims about a fuzzer's superiority at finding bugs.
Curiously enough, however, we find no strong agreement on which fuzzer is superior if we compared multiple fuzzers in terms of coverage achieved instead of the number of bugs found. The fuzzer best at achieving coverage, may not be best at finding bugs.

Bio

Marcel Böhme leads the Software Security research group at the Max Planck Institute for Security and Privacy (MPI-SP) in Germany. Previously, he was a Senior Lecturer at Monash University in Australia and a PostDoc at the TSUNAMi Security Research Centre in Singapore and the CISPA-Helmholtz Zentrum in Germany. Marcel received his PhD from the National University of Singapore.
His current research interest is the automatic discovery of security flaws at the very large scale. One part of his group develops the probabilistic foundations of automatic software testing (i.e., finding bugs by generating executions) to elucidate fundamental limitations of existing techniques and to explore the assurances that software testing provides when no bugs are found. The other part of his group develops practical vulnerability discovery tools that are widely used in software security practice. For instance, Entropic is the default power schedule in LibFuzzer which powers the largest fuzzing platforms at Google and Microsoft, fuzzing hundreds of security-critical projects on 100k machines 24/7. His tools have discovered 100+ bugs in widely-used software systems, more than 70 of which are security-critical vulnerabilities registered as CVEs at the US National Vulnerability Database.

 

19.05.2022 – Improving development tool impact through natural developer interfaces

Speaker: Prof. Dr. Reid Holmes

Host: Prof. Dr. Thomas Fritz

Abstract

Software developers continually adapt how they work to the increasing complexity of modern software systems. Software development tools help developers to understand, plan, and perform their tasks on large systems more effectively than without tool support. While developers know that these tools can improve productivity, many tools go unused by developers despite their value. Three reasons for this are that developers often do not know which tools they should use, when they should use them, or how to apply them to evolve their systems. 
In this talk, I will discuss how natural developer interfaces offer an opportunity to more seamlessly support software evolution by better aligning tools with how developers work on their tasks in practice. The most famous example of a natural developer interface is autocomplete, which developers invoke continuously as it helps them in a way that aligns with their needs. By providing developers tools that naturally fit their workflows, tools can be more widely used and have greater impact on developer's tasks.

Bio

Reid Holmes is an Associate Professor of Computer Science at the University of British Columbia. He is the recipient of the 2018 CS-Can/Info-Can Young Computer Science Researcher Award, a national recognition for Canadian Computer Scientists who have made outstanding research contributions. He is also the recipient of five ACM Distinguished Paper Awards and was runner-up for the 2015 ICSE Test-of-Time Award. His research focuses on deepening our understanding of how developers work with large, long-lived codebases and developing novel approaches to help them effectively evolve these systems while maintaining overall system quality. He has graduated over twenty students who have gone on to both academic and industrial positions.

 

02.06.2022 – Decision Support with User-Centered Visual Analytics

Speaker: Prof. Dr. Jörn Kohlhammer

Host: Prof. Dr. Jürgen Bernard

Abstract

Domain experts are increasingly convinced of the potential of visual analytics, the integration of automated analysis and visualization, for data-driven decision making as this research field moves from basic to applied research. At the same time, the rapidly increasing amount of data is a challenge in many application areas and practitioners are looking for more effective ways to analyze their data and communicate insights. We are working with experts in several application fields, including security analysts, engineers, doctors, and business analysts. In any of these domains, visual analytics solutions not only have to create user-, data- and task-centered visualizations: they use automated techniques and have to explain results to domain experts to allow a reasonable analytical discourse between human and visual analytics system. In this talk, I will discuss several examples from applied research projects with domain experts. My talk will also cover some of our current research topics in applied visual analytics looking at the future role of visual analytics in these domains.

Bio

Jörn Kohlhammer is head of the Competence Center for Information Visualization and Visual Analytics, and Professor for User-Centered Visual Analytics at TU Darmstadt. He has a Ph.D. from TU Darmstadt and an MSc from the Ludwig-Maximilian University in Munich. His competence center develops solutions for several application domains, including visual business analytics, medical data analysis of electronic health records, decision support in the public sector, and cybersecurity. Jörn is regular member of program committees for conferences like IEEE VIS and EuroVis, and acts as a reviewer for many conferences and journals. His personal research interests include user-centered design and decision-centered visualization.