CV Tag: Machine Learning

  • AI researcher Philipp Hennig receives ERC Consolidator Grant

    AI researcher Philipp Hennig receives ERC Consolidator Grant

    Philipp Hennig is Professor of Machine Learning Methods in the Department of Informatics at the University of Tübingen, part of Cyber Valley, and a member of the board of the Tübingen AI Center and the cluster of excellence Machine Learning: New Perspectives for Science. 

    ANUBIS objectives

    The ANUBIS project aims to manage the computing resources of scientific AI applications comprehensively and consistently. In climate models, geological and neuroscientific simulations, many inference problems occur in which information or a data set is used to indirectly deduce or “infer” the information that is actually being sought. "To solve such inference problems, we need large amounts of data and large computing resources," Hennig explains. "A single partial data set from climate research, for example, can be as large as 100 terabytes (i.e. 100,000 gigabytes)."

    Data processing challenges

    In the project, what a computer does – computing – is regarded as a source of information. This information relates to questions that cannot be answered definitively and perfectly (such as "What will the weather be like in 10 years' time?"). Such questions usually consist of an infinite number of sub-questions that are not fully answered by the calculation. Therefore, while the computer is answering these sub-questions, it should simultaneously keep a record of which parts of the question have already been answered and how well. This is not so easy, because the record-keeping itself is also a calculation. For the whole thing to remain feasible, the record-keeping must therefore be simpler than the calculation itself.

    Innovative approaches to solutions

    To do this, it will be necessary to extend modern machine learning methods so that their functionality harmonizes with the new concept. The advantages of the extended methodology are not only more economical algorithms, but also new functionality. "It will then be easier for researchers from the geosciences, climate sciences and neurosciences, for example, to flexibly incorporate very different types of data such as specific measurements, simulation data and expert knowledge into their code," says Hennig. "The uncertainty of the information from the different data sources must be quantified. Neither was possible with the previous methods."

    Related Links

    Media Contact

    Claudia Brusdeylins
    Tübingen AI Center / Universität Tübingen

  • Software Tutorial Series

    Software Tutorial Series

    The talk is open to everyone who uses python and is interested in sharing knowledge.

    The meeting link and login data will be published soon.

     

  • Software Tutorial Series

    Software Tutorial Series

    The lecture is about the benefits of organizing the code in packages (i.e. having your software pip friendly, unit tests friendly, automated html documentation generation friendly). Vincent Berenz will also talk about poetry (https://python-poetry.org/) as a convenient way to manage virtual environments and dependencies.

    The talk is open to everyone who uses python and is interested in sharing knowledge.

    Meeting link
    Meeting number: 2734 722 9158
    Password: R4xKxnCMr62

     

  • SimTech2023

    SimTech2023

     

    Vom 4. bis 6. Oktober 2023 laden die Universität Stuttgart und der Exzellenzcluster EXC 20275 "Data-integrated Simulation Science" (SimTech) führende Forscher im Bereich der datenintegrierten Simulationswissenschaft zur SimTech2023-Konferenz nach Stuttgart ein. Diese neue Art der Simulation integriert verschiedene Arten von Daten, von experimentellen Daten über Simulationsergebnisse bis hin zu Metadaten, um die Effizienz und Zuverlässigkeit von Simulationen zu verbessern.

     

    Conference Webpage

     

    Topics of the conference

    The conference will cover topics such as

    • classical computational science enhanced by learning based components,
    • data-based scale bridging,
    • high performance computing,
    • numerical solvers optimized by meta-data based methods,
    • machine learning for simulation and visualization,
    • data-based (surrogate) modeling for interactive applications and control problems,
    • Bayesian approaches for uncertainty reduction and model assessment and many more.

     

    Is this conference for you?

    The conference is intended to be a major platform for promoting exchange amongst representatives of the international research community in these interdisciplinary fields, bringing together mathematicians and life scientists, engineers and chemists, computer scientists and sociologists, physicists and philosophers. In keynote lectures, renowned scientists present the latest ideas at the interface between data and simulation. With this multi-disciplinary environment, we continue to advance towards our goal at the Stuttgart Cluster of Excellence SimTech: From Multi-X Simulations and Data-Driven Approaches to Data-Integrated Simulation Science.

    We are looking forward to welcome you all in Stuttgart!

     

    Organizer:

    Professor Miriam Schulte

  • Cluster Colloquium “Machine Learning”

    To inform the design of artificial systems, we are investigating how the fly, Drosophila melanogaster, selects and controls its behaviors and how this process can be modulated by learning. Prof. Pavan Ramdya will discuss how we are pursuing this research program using a combination of machine learning, microscopy, and modeling.

     

    The Speaker

    Prof. Pavan Ramdya, Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland & Director of the Neuroengineering Laboratory Website

     

    The Event

    This lecture will take place on site in the lecture hall of Maria-von-Linden-Straße 6 (EG) and will also be broadcasted on Zoom.

     

    Seminar Series of the Cluster of Excellence "Machine Learning"

    This seminar is part of the Seminar Series of the Cluster of Excellence "Machine Learning", University of Tübingen.
    Wednesdays  | 2.00 – 3.00 pm | followed by a get-together.

     

  • Master class: The Numerics of Machine Learning

    Master class: The Numerics of Machine Learning

    A new course is available on the YouTube channel “Tübingen Machine Learning”. The lecture series explains the view of probabilistic numerics, which assumes that computation itself is an inference process.

    Materials:

    • To the slides (github). All slides come with embedded, time-stamped links to these youtube videos (click on the icon on the bottom right of each slide).
    • To the playlist (youtube). Videos are uploaded in semantic groups over the course of January and February 2023.

    The introductory lecture of the course will be given by Philipp Hennig, Cyber Valley co-spokesperson and professor of machine learning methods at the Tübingen AI Center of the University of Tübingen. The course on machine learning numerics will be taught in the winter semester 2022/23 as part of the master's program in Machine Learning.

    Freely accessible learning and teaching videos

    The new series of lectures is one of several available for free on the YouTube channel “Tübingen Machine Learning” which is the result of a collaboration between the cluster's faculty and the Tübingen AI Center. “We started this during Corona, and have since built up a broad audience” says Philipp Hennig. For those interested in learning about machine learning without prior knowledge, he recommends the course series “Introduction to Machine Learning” by Dmitry Kobak. Those who want to stay up to date on current research can watch the three-minute “Research Spotlight Videos” that researchers publish on new scientific papers. The channel, which has more than 25,000 subscribers, is an example of OER practice in action in the Cyber Valley ecosystem. OER stands for Open Educational Resources, which are freely accessible learning and teaching materials that enable people around the world to further their education in the field of machine learning. 

    To the overview of the lecture series 

  • Research Talk: Semantic Probabilistic Layers for Neuro-Symbolic Learning

    Extending current deep learning architectures to correctly conform to prior knowledge and capture symbolic constraints while guaranteeing their satisfaction is a big open problem. Doing so would benefit many structured-output prediction (SOP) tasks in machine learning such as predicting rankings, hierarchies and other structured objects. In this talk, I will discuss how we can design a predictive layer for SOP that guarantees its predictions to be consistent with a set of predefined symbolic constraints. This Semantic Probabilistic Layer (SPL) can act as a drop-in replacement of existing predictive layers while still being amenable to end-to-end learning via exact maximum likelihood. SPLs combine exact probabilistic inference with logical reasoning in a clean and modular way via probabilistic circuits. We empirically demonstrate that SPLs outperform other neuro-symbolic approaches in terms of accuracy on challenging SOP tasks including hierarchical multi-label classification, pathfinding and preference learning, while retaining perfect constraint satisfaction.

     

    Antonio Vergari is a Lecturer (Assistant Professor) in Machine Learning at the University of Edinburgh. His research focuses on efficient and reliable machine learning in the wild; tractable probabilistic modeling and combining learning with complex reasoning. He recently was awarded an ERC Starting Grant on automating probabilistic reasoning for trustworthy ML. Previously he was postdoc in the StarAI Lab lead by Guy Van den Broeck at UCLA. Before that he did a postdoc at the Max Planck Institute for Intelligent Systems in Tuebingen in the Empirical Inference Department of Bernhard Schoelkopf. He obtained a PhD in Computer Science and Mathematics at the University of Bari, Italy. He likes to tease and challenge the probabilistic ML community at large on how we desperately need reliable ML an AI models nowadays. To this extent, he organized a series of tutorials, workshops, seminars and events at top ML and AI venues such as UAI, ICML, AAAI, IJCAI and NeurIPS and last year a Dagstuhl Seminar.

  • Great success at NeurIPS 2022

    Great success at NeurIPS 2022

    More than 2,900 papers have successfully passed the review process at NeurIPS. As in previous years, this includes researchers from the Max Planck Institute for Intelligent Systems, the University of Tübingen, the Max Planck Institute for Biological Cybernetics, the University of Stuttgart, the Bosch Center for Artifical Intelligence, Mercedes-Benz Innovation and Amazon Web Services in Tübingen. A complete list of papers, authors and their affiliations is available here.

    NeurIPS is one of the three main conferences with high impact on machine learning and AI, along with the International Conference on Machine Learning (ICML) and the International Conference on Learning Representations (ICLR). The conference was first held in 1987 and is now a multi-level interdisciplinary annual conference that includes invited talks, demonstrations, symposia, and oral and poster presentations of papers that successfully emerge from a rigorous review process. In addition to the conference, there is a technical exhibition focused on machine learning in practice, a series of tutorials, and topical workshops that provide a less formal setting for the exchange of ideas.