© pixabay
Wir, das Institute of AI in Management an der LMU München, sind begeistert von KI und den dynamischen Entwicklungen in diesem Bereich. Deshalb möchten wir Einblicke aus erster Hand in die aktuelle Forschungsarbeit von angesehenen Wissenschaftler:innen aus der ganzen Welt, geben. Wir freuen uns jedes Semester großartige Gastredner:innen für unsere Vortragsreihe gewinnen zu können.
Alle Vorträge finden online statt und sind per Zoom erreichbar für jeden, der Interesse hat. Unser Ziel ist es einen Überblick über aktuelle Trends in der KI Forschung zu geben. Die Vorträge finden immer am Donnerstag statt und bestehen aus ca. 45-60 Minuten Präsentation, gefolgt von Diskussion, Feedback und Q&A. Wir freuen uns darauf, Sie herzlich begrüßen zu dürfen.
Alle Informationen zu Terminen, Gastredner:innen und ihren Themen, inkl. Zoom Links werden im Laufe der Zeit auf dieser Veranstaltungsseite veröffentlicht.
Sie sind herzlich eingeladen sich für unseren Newsletter anzumelden, über den wir alle kommenden Veranstaltungen dieser Serie kommunizieren. Hier geht es zu unserer Anmeldeseite.
Die Terminserie ist eine gemeinsame Initiative mit Partnern von führenden nationalen Universitäten, unter der Leitung von Prof. Stefan Feuerriegel, LMU München:
- Prof. Markus Weinmann, University of Cologne
- Prof. Stefan Lessmann, Humboldt University Berlin
- Prof. Mathias Kraus, Friedrich-Alexander University Erlangen-Nuremberg
- Prof. Niklas Kühl, University of Bayreuth
- Dr. Michael Vössing, Karlsruhe Institute of Technology
- Prof. Oliver Müller, University of Paderborn
- Prof. Nicolas Pröllochs, Justus-Liebig-University Gießen
- Prof. Christian Janiesch, TU Dortmund
- Prof. Gunther Gust, University of Würzburg
- Prof. Tobias Brandt, University of Münster
- Prof. Yash Raj Shrestha, University of Lausanne
- Prof. Burkhardt Funk, Leuphana University Lüneburg
- Prof. Nadja Klein, TU Dortmund
- Prof. Martin Spindler, University of Hamburg
- Prof. Niki Kilbertus, TU Munich
- Prof. Stefan Bauer, TU Munich
- Prof. Henner Gimpel, University of Hohenheim
- Prof. Alexander Benlian, TU Darmstadt
- Prof. Oliver Hinz, Goethe University, Frankfurt
- Prof. Ekaterina Jussupow, TU Darmstadt
- Prof. Anne-Sophie Mayer, Vrije Universiteit Amsterdam
Do. 20.03.2025
Gastredner: Alejandro Schuler, Division of Biostatistics, UC Berkeley
Präsentation: RieszBoost: Gradient Boosting for Riesz Regression
Abstract: Answering causal questions often involves estimating linear functionals of conditional expectations, such as the average treatment effect or the effect of a longitudinal modified treatment policy. By the Riesz representation theorem, these functionals can be expressed as the expected product of the conditional expectation of the outcome and the Riesz representer, a key component in doubly robust estimation methods. Traditionally, the Riesz representer is estimated indirectly by deriving its explicit analytical form, estimating its components, and substituting these estimates into the known form (e.g., the inverse propensity score). However, deriving or estimating the analytical form can be challenging, and substitution methods are often sensitive to practical positivity violations, leading to higher variance and wider confidence intervals. In this paper, we propose a novel gradient boosting algorithm to directly estimate the Riesz representer without requiring its explicit analytical form. This method is particularly suited for tabular data, offering a flexible, nonparametric, and computationally efficient alternative to existing methods for Riesz regression. Through simulation studies, we demonstrate that our algorithm performs on par with or better than indirect estimation techniques across a range of functionals, providing a user-friendly and robust solution for estimating causal quantities.
Quelle: https://arxiv.org/abs/2501.04871
Uhrzeit: 16:00 CET
Sprache: Englisch
Vergangene Vorträge:
Do. 13.02.2025
Gastredner: Alex Luedtke, Department of Statistics, University of Washington
Präsentation: Simplifying Debiased Inference via Automatic Differentiation and Probabilistic Programming
Abstract: The speaker would introduce an algorithm that simplifies the construction of efficient estimators, making them accessible to a broader audience. 'Dimple' takes as input computer code representing a parameter of interest and outputs an efficient estimator. Unlike standard approaches, it does not require users to derive a functional derivative known as the efficient influence function. Dimple avoids this task by applying automatic differentiation to the statistical functional of interest. Doing so requires expressing this functional as a composition of primitives satisfying a novel differentiability condition. Dimple also uses this composition to determine the nuisances it must estimate. In software, primitives can be implemented independently of one another and reused across different estimation problems. The speaker provides a proof-of-concept Python implementation and showcase through examples how it allows users to go from parameter specification to efficient estimation with just a few lines of code.
Quelle: arXiv:2405.08675
Sprache: Englisch
Do. 09.01.2025
Gastredner: Chengchun Shi, Department of Statistics, London School of Economics and Political Science
Präsentation: Experimental Designs for A/B Testing in Marketplaces
Quelle: https://arxiv.org/pdf/2408.05342
Sprache: Englisch
Do. 12.12.2024
Gastredner: Mats Julius Stensrud, Chair of Biostatistics, Swiss Federal Institute of Technology Lausanne
Präsentation: On Optimal Treatment Regimes Assisted by Algorithms
Abstract: Decision makers desire to implement decision rules that, when applied to individuals in the population of interest, yield the best possible outcomes. For example, the current focus on precision medicine reflects the search for individualized treatment decisions, adapted to a patient's characteristics. In this presentation, the speaker will consider how to formulate, choose and estimate effects that guide individualized treatment decisions. In particular, he will introduce a class of regimes that are guaranteed to outperform conventional optimal regimes in settings with unmeasured confounding. He will further consider how to identify or bound these "superoptimal" regimes and their values. The performance of the superoptimal regimes will be illustrated in two examples from medicine and economics.
Sprache: Englisch
Do. 28.11.2024
Gastredner: Manoel Horta Ribeiro, Department of Computer Science, Princeton University (incoming)
Präsentation: Content Curation in Online Platforms
Abstract: Online platforms like Facebook, Wikipedia, Amazon, and Linkedin are embedded in the very fabric of our society. They “curate content”, moderate, recommend, and monetize it, and, in doing so, can impact people’s lives positively or negatively. In this talk, the sepaker will highlight the need to go beyond how these curation practices are currently designed and tested. He will argue that academic research can and should guide policy and best practices by discussing two projects he worked on during his doctorate. In the first project, the speaker will describe a large natural experiment on Facebook that allowed measuring the causal effect of removing rule-breaking comments on users’ subsequent behavior. In the second project, he will present results on the efficacy of “deplatforming” Parler, a large social media website, on its users’ information diets. Finally, the speaker will discuss future research directions on improving online platforms, emphasizing the opportunities and challenges posed by the popularization of generative AI. Altogether, the work of the speaker indicates that we can improve online platforms—and, by extension, our lives—if we rigorously investigate the causal effect of content curation practices.
Sprache: Englisch