Agenda (2023)

AGENDA (2023)

We are proud to have presented an impressive range of talks! The keynotes and technical talks were given by experienced data scientists from both academia and industry. The program was completed by a poster session featuring promising students who wanted to share their latest data science projects with the audience.

Morning Session (10 am – 1 pm CET)
(click to expand the full schedule)

10:00 am: Opening Remarks

10:30 am: Explaining the story behind data with Causal Inference by Clara Cullmann-Petroll (e:fs TechHub GmbH)

Structural causal models (SCMs) and causal inference offer a transparent way to model data-generating processes. Together with counterfactual queries, it is possible to explain complex systems. Learning SCMs from purely observational data has theoretical limitations and to obtain a more reliable model, both expert knowledge and interventional data can be incorporated. This talk focuses on an industry point of view on modeling SCMs, causal inference methods and how they can be applied in real-world problems in order to ensure the safety of systems.

11:00 am: Revolutionizing Design of Experiments using Bayesian Optimization by Dr. Lavinia Israel (ams OSRAM)

Bayesian optimization is a sequential strategy to optimize objective functions that are unknown and expensive to evaluate. It builds a surrogate model – usually a Gaussian process – for the objective and quantifies the uncertainty in that surrogate. It then uses an acquisition function defined from this surrogate model to decide where to evaluate next. At ams OSRAM we have many processes where process parameters need to be optimized to improve quality and efficiency. These processes do not have a functional representation and in general evaluations (= measurements) are expensive and time-consuming. The well-established method Design of Experiments has some drawbacks which we try to overcome by using Bayesian Optimization. In first projects we showed that we achieve optimal results with very few measurements.

11:30 am: Coffee break

11:45 am: Easily build machine learning applications using Gradio by Merve Noyan (Hugging Face)

Libraries like Gradio make it very easy to build UIs and serve it in minutes. It’s a library one can use to build interfaces to their data or ML apps using already existing input-output components and more. It reduces workload of data professionals that want to showcase their work with a neat UI, leaving them more time to improve their model or analysis. In this demo, I will walk the audience through how to easily build and demo machine learning applications using Gradio.

12:30 pm: Possibilities and limits of telepresence robotics and teletherapy in the context of the digital divide in society by Prof. Dr. Sonja Haug (OTH Regensburg)

The digital divide means that parts of the population are more or less included or excluded when areas of society become increasingly digitized. In the healthcare sector, digitization has not progressed very far in Germany; however, digitization is currently being increasingly discussed and promoted here as well. The paper presents results of a study in which patients with stroke test the possibility of additional care and therapy applications with a telepresence robot and therapy apps. Measures of technology affinity, technology competence, technology use, and technology acceptance among elderly and stroke patients are presented. Comparative information is also provided on other stakeholders such as relatives and therapy staff. The digital divide is considered based on differences in technology acceptance by age, gender, and education level. Chances and limitations of telepresence robotics and teletherapy are discussed against the background of societal developments, such as urban-rural differences.

1:00 pm: Lunch break

Afternoon Session (1:30 pm – 5:30 pm CET)
(click to expand the full schedule)

1:30 pm: Poster session

2:30 pm: Trustworthy and robust AI by Dr. Mojdeh Golagah (Infineon)

With the prevalent application of AI in smart home systems, medical diagnostic, and automotive industry, it is crucial for these systems to be trustworthy. Trustworthiness has important aspects including robustness and generalization. Robustness refers to the ability of a system to deal with execution errors, erroneous inputs, or unseen data. Generalization represents the capability to extract knowledge from limited training data to make accurate predictions regarding unseen data in an application environment. AI practitioners consider system performance such as accuracy to be the main metric in their workflows. However, this metric is not sufficient to reflect the trustworthiness of AI systems. Lack of trustworthiness violates customers’ expectations and harms the brand.  In our presentation, I will introduce techniques to evaluate and improve robustness of AI applications.

3:00 pm: Open Web Search and its Legal Implications by Dr. Jelena Mitrović (University of Passau)

Since September 2022, University of Passau and its chair of Data Science are coordinating the EU project, the first project the EU has funded to get tomorrow’s web search up and running. 14 renowned European research and computing centers have joined forces to develop an open European infrastructure for web search. The initiative will be contributing to Europe’s digital sovereignty as well as promoting an open, human-centered search engine market. In this talk, we will describe the main aspects of this innovative project, focusing on its legal implications.

3:30 pm: Coffee break

3:45 pm: Ready for AI? by Dr. Olga Mordvinova ( GmbH)

Ready for AI? AI has been a familiar topic in science for a long time and is now also a familiar subject in the business world. But how about its application in industrial reality? Let’s talk about what are hurdles and challenges and what benefits are already tangible.

4:30 pm: What’s the big deal about OpenAI? by Anupma Raj (Microsoft)

OpenAI claims to pioneer a new era of AI innovation. But what is behind the hype? Let’s talk about the service and how you can combine OpenAI’s language models to build secure and reliable solutions which uphold industry-leading Responsible AI practices.

5:15 pm: Closing remarks

5:30 – open end: Get together

No program, just delicious food and plenty of networking opportunity.