Cognitive Modeling

Matthias Karlbauer

Cognitive Modeling
Wilhelm Schickard Institute for Computer Science
University of Tübingen

Scholar of the International Max Planck Research School for Intelligent Systems (IMPRS-IS)
Funded by the cluster of excellence Machine Learning: New Perspectives for Science

Room C 421
Sand 14
72076 Tübingen, Germany

Phone: +49 7071 29 70543
Email: matthias.karlbauer@uni-tuebingen.de


Research interests

  • Machine learning and deep learning
    • Generative neural networks
    • Recurrent neural networks
    • Convolutional neural networks
    • Physics aware neural networks
  • Spatiotemporal data prediction
    • Weather forecasting, focusing on global and regional temperature and geopotential prediction
    • Determining links between modeling cognitive and physical processes
  • Sustainability and climate protection

Biographical information

  • Since March 2019: Doctoral student in the Cognitive Modeling group, Eberhard Karls University of Tuebingen and scholar of the International Max Planck Research School for Intelligent Systems (IMPRS-IS) program
  • 2015 - 2018: Master of Cogntitive Science, Eberhard Karls University of Tübingen
  • 2012 - 2015: Bachelor of Cognitive Science, Eberhard Karls University of Tübingen

Supervision

2022

  • BSc Emil Breustedt (2022), Contrasting Traditional and Learned Super-Resolution Methods on Weather Maps over Germany.
  • Lab project Cosku-Can Horuz (2022), Inferring Unknown Boundary Conditions of Spatiotemporal Advection-Diffusion Processes via Finite Volume Neural Network. Co-supervised with Sebastian Otte.
  • BSc Andreas Schaible (2022), Exploration eines Physik-Motivierten GNNs auf Raumzeitlich Heterogenen Observationsdaten.
  • MSc Adrian Stock (2022), Modeling Weather Station Data with Graph Neural Networks. Co-supervised with Tobias Menge and Jannik Thümmel.
  • BSc Adrian Welter (2022), Applying DISTANA to Sequential Three-Dimensional Fluid Flow Data.

2021

  • BSc Magnus Kaut (2021), Implementing DISTANA in PyTorch Geometric and Deep Graph Library.
  • BSc Linda Ulmer (2021), Time-Constrained Active Tuning in Recurrent Neural Networks. Co-supervised with Sebastian Otte.
  • BSc Arthur Otte (2021), Analysis and Extension of a Neural Network Model for Spatio-Temporal Prediction.
  • MSc Lars Gehrke (2021), GPU Implementation of the Graph Neural Network DISTANA with CUDA.

2020

  • BSc Lina Gönnheimer (2020), Predicting a 2D Wave Pattern with an Echo State Network.
  • MSc Tobias Menge (2020), Implementation and Evaluation of REPRISE for DISTANA regarding Static Context Inference for Wave Prediction.

2019

  • MSc Andreas Schmitt (2019), Exploring the Capabilities of a Spatio-Temporal, RNN-based Architecture using Simulated Fluid Data and Considering Contextual Information.
  • Lab project Leon Varga (2019), Introduce Sparse Information Processing in a Dense Grid by using REPRISE.

Teaching

2022

  • Lecture: Generative and Recurrent Neural Networks. Tutoring and holding lecture about Graph Neural Networks.

2021

  • Teamproject: Raumzeitliche Wetterdatenvisualisierung. Student supervision.
  • Lecture: Generative and Recurrent Neural Networks. Tutoring and holding lecture about Graph Neural Networks.

2020

  • Lecture: Advanced Artificial Neural Networks. Tutoring.
  • Lecture: Kognitive Architekturen. Tutoring.

2019

  • Lecture: Advanced Artificial Neural Networks. Tutoring.
  • Lecture: Kognitive Architekturen. Tutoring.

Publications

2022

  • Matthias Karlbauer, Timothy Praditia, Sebastian Otte, Sergey Oladyshkin, Wolfgang Nowak, and Martin V. Butz (2022). Composing Partial Differential Equations with Physics-Aware Neural Networks. Submitted to ICML.

2021

  • Matthias Karlbauer, Tobias Menge, Sebastian Otte, Hendrik P. A. Lensch, Thomas Scholten, Volker Wulfmeyer, and Martin V. Butz (2021). Latent State Inference in a Proceedings of the International Conference on Artificial Neural Networks (ICANN), 384-395.
  • Melvin Ciurletti, Manuel Traub, Matthias Karlbauer, Martin V. Butz, and Sebastian Otte (2021). Signal Denoising with Recurrent Spiking Neural Networks and Active Tuning. Proceedings of the International Conference on Artificial Neural Networks (ICANN), 220-232.
  • Timothy Praditia, Matthias Karlbauer, Sebastian Otte, Sergey Oladyshkin, Martin V. Butz, and Wolfgang Nowak (2021). Finite Volume Neural Network: Modeling Subsurface Contaminant Transport. ICLR workshop on Deep Learning for Simulation (SimDL).

2020

  • Sebastian Otte, Matthias Karlbauer, and Martin V. Butz (2020). Active Tuning. https://arxiv.org/abs/2010.03958.
  • Matthias Karlbauer, Sebastian Otte, Hendrik Lensch, Thomas Scholten, Volker Wulfmeyer, and Martin Butz (2020). Inferring, Predicting, and Denoising Causal Wave Dynamics. Proceedings of the International Conference on Artificial Neural Networks (ICANN), 566-577.
  • Matthias Karlbauer, Sebastian Otte, Hendrik Lensch, Thomas Scholten, Volker Wulfmeyer, and Martin Butz (2020). A Distributed Neural Network Architecture for Robust Non-Linear Spatio-Temporal Prediction. Proceedings of the European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN).

2018

  • Ruben Ellinghaus, Matthias Karlbauer, Karin M. Bausenhart, and Rolf Ulrich (2018). On the time-course of automatic response activation in the Simon task. Psychological research, 1-10.