Responsibilities
Research Group Leader (Tenure Track)
Contact
Computational Health Center
Email:
steffen.schneider@helmholtz-munich.de
Website:
https://dynamical-inference.ai/
Further Information
Keywords:
dynamical systems, representation learning, neural dynamics, identifiability theory, self-supervised learning, system identification
Brief research description:
We develop machine learning algorithms for representation learning and inference of nonlinear system dynamics, study how large and multi-modal biological datasets can be compressed into foundation models, and study their mechanistic interpretability. An up-to-date description of our research and latest news about publications and preprints is always available on our website: https://dynamical-inference.ai/
Selected publications:
Self-supervised contrastive learning performs non-linear system identification. arXiv, 2024. Rodrigo Gonzalez Laiz*, Tobias Schmidt*, and Steffen Schneider. https://arxiv.org/abs/2410.14673
Neuro-musculoskeletal modeling reveals muscle-level neural dynamics of adaptive learning in sensorimotor cortex. bioRxiv, 2024. Travis DeWolf*, Steffen Schneider*, Paul Soubiran, Adrian Roggenbach, and Mackenzie W Mathis. https://doi.org/10.1101/2024.09.11.612513
Learnable latent embeddings for joint behavioural and neural analysis. Nature, 2023. Steffen Schneider*, Jin Hwa Lee*, and Mackenzie W. Mathis. https://doi.org/10.1038/s41586-023-06031-6
Contrastive Learning Inverts the Data Generating Process. International Conference on Machine Learning (ICML), 2021. Roland Zimmermann*, Yash Sharma*, Steffen Schneider*, Matthias Bethge, and Wieland Brendel. https://doi.org/10.48550/arXiv.2102.08850 wav2vec:
Unsupervised Pre-training for Speech Recognition. Interspeech, 2019. Steffen Schneider, Alexei Baevski, Ronan Collobert, Michael Auli. https://doi.org/10.48550/arXiv.1904.05862