From April 10 to 12, the researchers of the Hasso Plattner Institute in Potsdam, Germany and invited speakers present their current research topics, providing a forum for exchange and discussion of ideas among students, industry and academia.

More details here.

Prof. Dr. Felix Naumann

The meeting will be dedicated to organizational issues of our cluster, including:

-              New members (DSE Research School and member groups)

-              Trip report(s): UCI visit

-              Weekly meetings: Topics

-              Spring report

-              Website

-              Shared tasks

-              I like / I wish

Prof. Sebastian Link

Title: Synthesizing Relational Database Schemata that Minimize Integrity Maintenance and Update Overheads

Abstract: Classical normalization of relational database schemata generates a lossless, dependency-preserving decomposition into Third Normal Form (3NF), that is in Boyce-Codd Normal Form whenever possible. In this talk, we will analyze the idea of parameterizing 3NF schemata by the numbers of minimal keys and non-key functional dependencies they exhibit. Conceptually, these parameters quantify, already at schema design time, the effort necessary to maintain data integrity, and allow us to break ties between 3NF schemata. Computationally, the parameters enable us to optimize normalization into 3NF according to different strategies we target. Operationally, we demonstrate through experiments that our optimizations translate from the logical level into significantly smaller update overheads during integrity maintenance. Hence, our framework provides access to parameters that guide the computation of logical schema designs which reduce operational overheads.

 Public holiday: Labour Day

Prof. Dr. Felix Naumann

Title: Conflicts of Interest and Reviewer Assignment

Abstract: Conflicts of interest in research occur when a person reviewing another person's work (paper, thesis, funding proposal, application, nomination, etc.) is potentially biased or can be perceived to be biased. Together, we will discuss different definitions of Col, the need and difficulty to disclose conflicts, how unbiased reviewing is subverted and what we can do about it. Please prepare for this meeting by searching and noting the Col-policies of the top conference in your particular field.

Prof. James Landay

Title: “AI For Good” Isn’t Good Enough: A Call for Human-Centered AI

Abstract: In this talk, Professor James Landay elaborates on his argument for an authentic Human-Centered AI. User-centered design integrates techniques that consider the needs and abilities of end users, while also improving designs through iterative user testing. Community-centered design engages communities in the early stages of design through participatory techniques. Societally-centered design forecasts and mediates potential impacts on a societal level throughout a project. Successful Human-Centered AI requires the early engagement of multidisciplinary teams beyond technologists, including experts in design, the social sciences and humanities, and domains of interest such as medicine or law, as well as community members.

Sedir Mohammed & Sebastian Schmidl

Title: Experiment Data Management

Abstract: As computer science researchers, we develop software systems to solve problems and execute many experiments to test and evaluate our systems. These experiments use different versions of our software, evaluate the software on many datasets, explore various settings, generate intermediate data, and hopefully produce results. Our experiments, thus, create massive amounts of experimental data, which needs to be stored and managed. In this mess, we may quickly lose track of particular results, their configuration, and whether the results are still up-to-date. How do we effectively manage our experimental data? In this talk, we propose a way to manage and query our experimental results using database technology. Based on two example setups, we show the benefits (and drawbacks) of storing experimental data within a database, such as querying using SQL, storage efficiency, access synchronization, and, in general, a better overview of all experiments and their history.

Prof. Dr. Ralf Herbrich

Title: How to Make the Most of a Conference Visit 

Abstract: In this discussion, I will share my experiences of both the typical structure of a scientific conference (week), what are my goals when attending a conference and the implications on what to do before, during and after the conference. I will dive deep into all elements of a scientific conference starting from tutorials, the technical program, poster session, workshops to the opening reception and dinners. The presentation shall only serve as a structure and initiation of a group discussion exchanging all our experiences in attending a conference.

Dr. Vasilis Ververis

Title: The Rise of Generative AI in Academic Writing: Challenges, Opportunities, and Ethical Considerations

Abstract: The advent of generative AI tools like ChatGPT has sparked a heated debate about their role in academic writing. This talk will delve into the implications of using these tools to write master theses and academic articles. We will explore the benefits of increased efficiency and the potential for enhanced creativity, as well as the challenges of maintaining academic integrity and the risks of plagiarism. The discussion will also touch on the ethical considerations surrounding the use of AI-generated content and the need for clear guidelines and policies to ensure responsible use in higher education. Join us as we navigate the complexities of integrating generative AI into the academic writing landscape.

Prof. Dr. Patrick Baudisch

Title: The Scientific Method

Ricardo Salazar Díaz 

Title: InferDB: In-Database Machine Learning Inference Using Indexes

Abstract: The performance of inference with machine learning (ML) models and its integration with analytical query processing have become critical bottlenecks for data analysis in many organizations. A ML inference pipeline typically consists of a preprocessing workflow followed by prediction with an ML model. Current approaches for in-database inference implement preprocessing operators and ML algorithms in the database either natively, by transpiling code to SQL, or by executing user-defined functions in guest languages such as Python. In this work, we present a radically different approach that approximates an end-to-end inference pipeline (preprocessing plus prediction) using a light-weight embedding that discretizes a carefully selected subset of the input features and an index that maps data points in the embedding space to aggregated predictions of an ML model. We replace a complex preprocessing workflow and model-based inference with a simple feature transformation and an index lookup. Our framework improves inference latency by several orders of magnitude while maintaining similar prediction accuracy compared to the pipeline it approximates.

Prof. Dr. Tilmann Rabl

Title: Proper Benchmarks and Evaluation

Abstract: In this session, we will discuss benchmarking and measurement. We will present some techniques for fast performance modelling and how to integrate this into the publication pipeline. Then we will further discuss terminology and some of the pitfalls in benchmarking and performance comparisons.

Martin Boissier, Robin van de Water, and Eshant English 

Title: Collaborations with the UCI and the living experience in Irvine

Abstract: Martin Boissier, Robin van de Water, and Eshant English, talk about their research stay at the UCI and life in Irvine. Additionally, Eshant introduces Conformal Prediction. Conformal prediction provides machine learning models with prediction sets that offer theoretical guarantees, but the underlying assumption of exchangeability limits its applicability to time series data. Furthermore, existing approaches struggle to handle multi-step ahead prediction tasks, where uncertainty estimates across multiple future time points are crucial. We propose JANET (Joint Adaptive predictioN-region Estimation for Time-series), a novel framework for constructing conformal prediction regions that are valid for both univariate and multivariate time series. JANET generalises the inductive conformal framework and efficiently produces joint prediction regions with controlled K-familywise error rates, enabling flexible adaptation to specific application needs. Our empirical evaluation demonstrates JANET's superior performance in multi-step prediction tasks across diverse time series datasets, highlighting its potential for reliable and interpretable uncertainty quantification in sequential data.

Prof. Dr. Gerard de Melo

Title: It's about Time! Time Management and Goals for Your Research

Abstract: Many researchers feel overwhelmed by the many tasks and duties that they are faced with on a daily basis. This encompasses paper deadlines, teaching duties, academic service, and of course also various personal commitments. We will discuss a few strategies to better manage your time, such as Allen's Getting-Things-Done approach and the Eat-that-Frog strategy, along with tools that one can use to support this.