Spring Term 2025

We are delighted to welcome a number of distinguished computer scientists to the upcoming IfI colloquium in the Spring Term FS 2025. We are looking forward to inspiring encounters with our guests, presenting topics from different areas of computer science.

The IfI colloquium is a free public event for researchers, students and the interested public and does not require registration. 

The IfI colloquium is held in English and takes place from 17:15 to 18:30 in room 2.A.01 at the Department of Informatics (IfI), Binzmühlestrasse 14, 8050 Zürich. 

All the talks are held onsite. Some of them might be streamed. Please check the website shortly before the respective presentation date.

If you have further questions please contact Karin Sigg.

Flyer to download (PDF, 244 KB)

* If you would like to get access to the talk please send an email until 16:00 on the day of the talk to  studies@ifi.uzh.ch

Newsletter IfI Colloquium
We announce our IfI Colloquium talk series every semester via email. If you want to subscribe to this mailing list please send an email to Karin Sigg.

27.02.2025 – Combining Reinforcement Learning and MPC - What does it tell us about model-based policies?

Speaker:  Prof. Dr. Sebastien Gros

Host: Prof. Dr. Davide Scaramuzza

Abstract

This talk aims at summarizing the results and insights presented in nearly 50 publications on the combination of Reinforcement Learning (RL) and Model Predictive Control (MPC). These investigations have shown how to achieve this combination, for which class of problems it is most useful, and where the difficulties are. In addition, they also provide rich insights into the use of MPC as a way of providing optimal control policies, especially when dealing with stochastic problems and economic objectives.

Bio

Sebastien Gros took his PhD degree in 2008 at the Automatic Control Lab, EPFL. After a bike trip in full autonomy from Switzerland to the Everest base camp, he worked in the wind industry in 2010-2011. He then joined the Optimal Control group at KU Leuven in 2011 as a postdoc where he worked on numerical optimization methods, and NMPC for complex mechanical applications. He then joined the University of Chalmers in 2013 as an Assistant Professor, where he worked on distributed optimization methods, autonomous driving, vehicle control and traffic optimization. He was promoted to Associate Professor in 2017. He joined the Dept. of Cybernetics at the Norwegian University of Technology (NTNU) in 2019 as a full Professor, and became head of Dept. in 2022. He has been working on learning methods for MPC since 2018.

06.03.2025 – Machine Learning-supported Network Management for User-centric Communication Networks

Speaker:  Prof. Dr. Michael Seufert

Host: Prof. Dr. Burkhard Stiller

Abstract

In modern user-centric network environments, ensuring a high Quality of Experience (QoE) for users of Internet applications is increasingly challenging due to the limitations of traditional network monitoring and management mechanisms. This talk will explore how machine learning approaches are emerging as powerful tools to address these challenges. Supervised machine learning models can demonstrate high accuracy in QoE monitoring even with encrypted network traffic, while reinforcement learning offers promising capabilities for proactive QoE management. Despite these advantages, we must also consider the limitations, including the risks of model overfitting and the requirement for labeled QoE data.

Bio

Michael Seufert is a Full Professor with the University of Augsburg, Germany, heading the Chair of Networked Embedded Systems and Communication Systems since 2023. He has been with the Chair of Communication Networks, University of Würzburg, Germany from 2019-2023 and from 2013-2017, with AIT Austrian Institute of Technology from 2018-2019, and with FTW Telecommunication Research Center Vienna from 2012-2013. He received the Diploma and Ph.D. degrees in computer science, the bachelor’s degree in economathematics, and the Habilitation degree in computer science from the University of Würzburg, Germany, in 2011, 2017, 2018, and 2023, respectively. He also received the First State Examination degree in mathematics, computer science, and education for teaching in secondary schools, in 2011. His research focuses on user-centric communication networks, including QoE of internet applications, AI/ML for QoE-aware network management, as well as group-based communications. He has published more than 100 papers in renowned international journals, conference, and workshops, and has been awarded multiple times by the international research community, winning, among others, 6 Best Paper Awards, 3 Best Demonstration Awards, and 2 Best PhD Thesis Awards. Since 2022, he is a Fellow of the prestigious DFG Emmy Noether Program and he is the recipient of the 2024 IEEE Communications Society (ComSoc) Technical Committee on Network Operation and Management (CNOM) Young Professional Award.

03.04.2025 – Building AI-Powered Apps: Challenges, Lessons, and Future Research Directions

Speaker:  Prof. Dr. Christopher J. Parnin

Host: Prof.  Dr. Thomas Fritz

Abstract

The integration of advanced AI, particularly Large Language Models (LLMs), into software products is transforming how users interact with technology and how software is built. Moving beyond button clicks, a user can now expect to communicate with software with natural language messages such as “I have extra vacation days to use up. Go ahead and create an OOF event for every Friday without any scheduled team meetings for every remaining vacation day”—and expect it to work. Traditional software architectures were not designed to handle such open-ended user interactions, introducing unprecedented challenges for software engineers and their development processes. In this talk, drawing on interviews with over 50 practitioners and a survey of 332 professionals, we discuss the pain points, challenges, and emerging solutions for overcoming these challenges. Additionally, we connect these challenges to experiences building product Copilot at Microsoft, including Visual Studio and Office products. Finally, we discuss the many challenges associated with evaluating AI-powered software, including the role of LLMs-as-a-judge in task evaluation. This colloquium offers a comprehensive overview of the challenges, research opportunities, and actionable strategies for shaping the future of robust, user-centric AI-powered systems.

Bio

Dr. Chris Parnin’s research spans developer productivity, cognition and learning, and automated infrastructure. He has published over 90 papers, received five SIGSOFT Distinguished Paper Awards, a 10-Year Impact Paper Award, a Google Faculty Award, and an NSF CAREER Award. His work has been featured in hundreds of international news articles, magazines, and frequently discussed in industry forums.
Currently, Dr. Parnin leads research efforts at Microsoft focused on building and evaluating Copilots for Visual Studio and the Office Suite. His experience also includes serving as a tenured professor at NC State University, contributing to Microsoft Research's Human Interactions in Programming group, conducting field studies with ABB Research, and over a decade of professional programming in the defense industry.
Dr. Parnin has collaborated with industry partners to bridge research and practice, including developing mentorship programs for Stack Overflow, creating anxiety-reducing features for CoderPad, and advancing equitable hiring solutions with Byteboard. He also co-organized five Continuous Deployment Summits, hosted by leading companies such as Facebook, Netflix, Google, Microsoft, and Twitter. These summits fostered collaboration and innovation in continuous deployment practices, resulting in impactful publications and educational resources for students and professionals.

10.04.2025 – Volumes meet Gradient-based Optimization

Speaker:  Prof. Dr. Rüdiger Westermann

Host: Prof.  Dr. Renato Pajarola

Abstract

Gradient-based optimization is a mathematical approach used to find the optimal solution to a problem by iteratively adjusting the decision variables based on the gradient of an objective function. It is widely used in engineering, machine learning, and various scientific disciplines due to its efficiency in handling high-dimensional and continuous optimization problems. In this talk I will discuss various applications of gradient-based optimization in scientific visualization, with the objective being either the generation of an accurate visual depiction of a volumetric scalar field, or the reconstruction of such a field from given images. I will demonstrate that the use of gradient-based optimization can reduce the memory and computational resources required in image synthesis tasks, and leads to advance computational performance and various applications in volume reconstruction tasks. Specifically, I will address super-resolution volume rendering and volume compression with  neural networks, cinematic volume rendering with differentiable 3D Gaussian splatting, and image-based volume reconstruction with automatic differentiation.

Bio

Rüdiger Westermann is a Professor for computer science at the Technical University of Munich. He is head of the chair for computer graphics and visualization. After his PhD in computer science he joined the computer graphics group at the University of Erlangen-Nuremberg as a research scientist. Before he became an assistant professor in the visualization group at the University of Stuttgart he was a research assistant in the mulitres group at Caltech and a visiting professor with the scientific computing laboratory at the University of Utah. In 2001 he was appointed by the RWTH-Aachen as an associate professor for scientific visualization in the department of computer science. Since 2003, Rüdiger Westermann is chair of the computer graphics and visualization group at TUM. He was honored an ERC Advanced Grant for research in the area 
of uncertainty visualization, and he served as a PI in the DFG funded center of excellence Waves2Weather. His current research interests include realtime graphics and interactive visual data analysis, topology optimization and GPU compute.

22.05.2025 – Enhancing Access: from Assistive Technology Development to Designing for User Acceptance

Speaker:  Prof. Khai Truong, Ph.D.

Host: Prof.  Elaine Huang, Ph.D.

Abstract

While assistive technologies have the potential to transform lives, many are abandoned after initial adoption. Why does this happen? In this talk, I will explore the gap between technological innovation and real-world user acceptance, focusing on the challenges designers and researchers face in creating solutions that are not only functional but also embraced by their users. I will present case studies that highlight the need to understand how psychosocial factors—such as stigma, identity, and social dynamics—shape users’ perceptions of assistive technology. Finally, I will discuss strategies for addressing these often-overlooked factors to enhance the value and acceptance of these technologies for the individuals they are meant to empower.

Bio

Khai Truong is a Professor in the Department of Computer Science at the University of Toronto. Khai received a Ph.D. degree in Computer Science and a Bachelor degree in Computer Engineering with highest honors from the Georgia Institute of Technology. He has been an active ubicomp researcher for over 20 years. His research interest lies at the intersection of human-computer interaction (HCI) and ubiquitous computing, and investigate tools and methods to support the development of novel ubiquitous computing systems and techniques and models to facilitate user interactions with off-the-desktop computing devices and services. His current work also includes the design and evaluation of assistive technologies and context sensing applications.
 

Click here for the archived IfI Colloquia pages.