Biomedical Sequence Data
The language of life is encoded as a sequence of amino acids in the form of proteins. The understanding of their properties is of fundamental interest both on the level of proteins for fundamental research as well as medical applications and on the level of peptides, i.e. protein fragments, due to applications in immunology. In particular, only for a tiny fraction of proteins is experimentally studied in enough detail such that its main functional properties are known. The work in the AML group at HHI focuses on the prediction of protein/peptide properties using the primary sequence alone, leveraging the recent advances in the field of natural language processing.
Medical Imaging Data
Medical imaging is the field of visual representation of biological structures in medicine. This makes it a very popular application for abstract algorithms. The different research methods of the AML group at HHI can be applied in an early stage such as the acquisition and reconstruction of medical images, as well in a later stage such as the diagnosis and detection of pathologies. Furthermore, the interpretable machine learning methods that we develop are of fundamental need for a transparent and safe deployment of machine learning algorithms.
Medical Time Series
Time series data constitutes a rich data modality that is highly relevant for applications in the medical domain as it includes for example electrocardiography (ECG), electrical impedance tomography (EIT), electroencephalography (EEG) but also gait data. The AML group develops algorithms for the analysis of such data with particular regard to quality criteria ranging from quantitative accuracy to interpretability and beyond.
Publications
Biomedical Sequence Data
[1] Nils Strodthoff, Patrick Wagner, Markus Wenzel, and Wojciech Samek: UDSMProt: Universal Deep Sequence Models for Protein Classification, Bioinformatics, 36(8):2401–2409, 2020.
[2] Johanna Vielhaben, Markus Wenzel, Wojciech Samek, and Nils Strodthoff: USMPep: Universal Sequence Models for Major Histocompatibility Complex Binding Affinity Prediction, BMC Bioinformatics 21, 279, 2020.
Medical Imaging Data
[1] Tatiana A. Bubba, Gitta Kutyniok, Matti Lassas, Maximilian März, Wojciech Samek, Samuli Siltanen, and Vignesh Srinivasan: Learning The Invisible: A Hybrid Deep Learning-Shearlet Framework for Limited Angle Computed Tomography, Inverse Problems, 35(6):064002, 2019.
[2] Miriam Hägele, Philipp Seegerer, Sebastian Lapuschkin, Michael Bockmayr, Wojciech Samek, Frederick Klauschen, Klaus-Robert Müller, and Alexander Binder: Resolving Challenges in Deep Learning-Based Analyses of Histopathological Images using Explanation Methods, Scientific Reports, 10:6423, 2020.
[3] Luis Oala, Cosmas Heiß, Jan Macdonald, Maximilian März, Wojciech Samek, and Gitta Kutyniok: Detecting Failure Modes in Image Reconstructions with Interval Neural Network Uncertainty, Proceedings of the ICML'20 Workshop on Uncertainty & Robustness in Deep Learning, 2020.
[4] Armin W. Thomas, Hauke R. Heekeren, Klaus-Robert Müller, and Wojciech Samek: Analyzing Neuroimaging Data Through Recurrent Deep Learning Models, Frontiers in Neuroscience, 13:1321, 2019.
Medical Time Series
[1] Nils Strodthoff, and Claas Strodthoff: Detecting and interpreting myocardial infarction using fully convolutional neural networks, Physiological Measurement 40, no. 1, 015001, 2019.
[2] Patrick Wagner, Nils Strodthoff, Ralf-Dieter Bousseljot, Dieter Kreiseler, Fatima I. Lunze, Wojciech Samek, and Tobias Schaeffter: PTB-XL, A Large Publicly Available Electrocardiography Dataset, Scientific Data, 7:154, 2020.
[3] Nils Strodthoff, Patrick Wagner, Tobias Schaeffter, and Wojciech Samek: Deep Learning for ECG Analysis: Benchmarks and Insights from PTB-XL, IEEE Journal of Biomedical and Health Informatics (to appear), 2020.
[4] Nils Strodthoff, Claas Strodthoff, Tobias Becher, Norbert Weiler, and Inéz Frerichs. Inferring respiratory and circulatory parameters from electrical impedance tomography with deep recurrent models. arXiv preprint 2010.09622, 2020.
[5] Irene Sturm, Sebastian Lapuschkin, Wojciech Samek, and Klaus-Robert Müller: Interpretable Deep Neural Networks for Single-Trial EEG Classification, Journal of Neuroscience Methods, 274:141–145, 2016.
[6] Fabian Horst, Sebastian Lapuschkin, Wojciech Samek, Klaus-Robert Müller, and Wolfgang I. Schöllhorn: Explaining the Unique Nature of Individual Gait Patterns with Deep Learning, Scientific Reports, 9:2391, 2019.
MCMC Physics
[1] Kim Nicoli, Shinichi Nakajima, Nils Strodthoff, Wojciech Samek, Klaus-Robert Müller, and Pan Kessel. Asymptotically Unbiased Estimation of Physical Observables with Neural Samplers. Phys. Rev. E 101, 023304, 2020.
[2] Kim Nicoli, Pan Kessel, Nils Strodthoff, Wojciech Samek, Klaus-Robert Müller, and Shinichi Nakajima. Comment on "Solving Statistical Mechanics Using VANs": Introducing saVANt - VANs Enhanced by Importance and MCMC Sampling. arXiv preprint 1903.11048, 2019.
[3] Johanna Vielhaben, and Nils Strodthoff. Generative Neural Samplers for the Quantum Heisenberg Chain. arXiv preprint 2012.10264, 2020.
[4] Stefan Bluecher, Lukas Kades, Jan M. Pawlowski, Nils Strodthoff, Julian M. Urban. Towards Novel Insights in Lattice Field Theory with Explainable Machine Learning. Phys. Rev. D 101, 094507, 2020.