[Home]

Please add short paragraphs about each project, naming scientific question/target, application area, used methods, and main outcome. If given provide a link to the according webpage. Provide the name of the responsible person and a date.

Highlights

Most recent (almost) published results from FZJ related to ML&DL

• "Rank Selection in Non-negative Matrix Factorization: systematic comparison and a new MAD metric", accepted IJCNN 2019. This work by Laura Muzzarelli, Susanne Weis, Simon B. Eickhoff and Kaustubh R. Patil from INM-7 looks at the unsolved problem of how to estimate the rank of NNMF decomposition for a given dataset. In this work various existing and a novel method were compared on simulated and real world data.

Groups at FZJ, dealing with ML and DL

Here we provide a plain list of links to groups related to ML and DL. For more information, please visit the according group webpages. In order to find groups probably related to your research, simply search for keywords you're interested in.

• Big Data Analytics at INM-1
  Contact: Timo Dickscheid, t.dickscheid@fz-juelich.de
  image data, classification, semantic segmentation, CNN, SVM, Capsule Networks, siamese networks, histology, HPC

• Applied Machine learning at INM-7
  Contact: Kaustubh R. Patil, k.patil@fz-juelich.de
  Data: neuroimaging (fMRI, DTI, T1), human behavior, clinical symptoms and disgnosis
  Methods: classification and regression (RF, SVM, DNN) , clustering (k-means, GMM, spectral), structured prediction (CRF)

• Biomarker Development at INM-7
  Contact: Juergen Dukart, j.dukart@fz-juelich.de
  Data: neuroimaging (fMRI, DTI, T1, PET, SPECT), sensor-based data, clinical symptoms and diagnosis
  Methods: SVM, Decision trees, Linear and Logistic regression, Bayesian classifiers

• Cognitive Neuroinformatics at INM-7
  Contact: Sarah Genon, s.genon@fz-juelich.de
  Data: neuroimaging (MRI, PET), psychometric data
  Methods: classification and regression (SVM, RVM), clustering (k-means) and factorization (NNMF, PCA)

• Brain Variability at INM-7
  Contact: Susanne Weis, s.weis@fz-juelich.de
  Data: neuroimaging (fMRI, T1), human behaviour, sex differences
  Methods: classification and regression (RF, SVM)

• Helmholtz Analytics Framework (at JSC, INM-1, INM-6, IEK-8, ICS-6, IBG-3)
  Contact: Björn Hagemeier, b.hagemeier@fz-juelich.de
  Data: various
  Methods: generic methods at big scale, parallelized and targeted at JSC's HPC systems
  Domains and use cases:

  • Earth System Modelling:
    • Terrestrial Monitoring and Forecasting (IBG-3)
    • Cloud and Solar Power Prediction (IEK-8)
  • Structural Biology: Hybrid Data Analysis and Integration for Structural Biology (JSC, ICS-6)
  • Neuroscience:
    • High-Throughput Image-Base Cohort Phenotyping (INM-1)
    • Multi-Scale Multi-Area Interaction in Cortical Networks (INM-6)

• Simulation Laboratory Neuroscience at JSC
  Contact: Kai Krajsek, k.krajsek@fz-juelich.de
  Supervised and Unsupervised Learning, Medical Image Analysis (e.g. Segmentation, Registration, Orientation Estimation, Dimension Reduction, Diffusion-weighted Imaging Modelling), Inverse Modelling, Learning to Learn, Structural Plasticity

• Simulation Laboratory Biology at JSC
  Contact: Olav Zimmermann, olav.zimmermann@fz-juelich.de
  Hybrid methods for local and global protein structure prediction (ML + atomistic MCMC simulation), pattern mining in peptide aggregation simulations
  Methods: classification/regression: SVM, SVR, SSVM, clustering/dimension reduction: Isomap, TICA, pymafia

• Simulation Laboratory Quantum Materials at JSC
  Contact: Edoardo Di Napoli, e.di.napoli@fz-juelich.de
  Applying and developing Machine Learning methods for the prediction of thermodynamic properties (Enthalpies of formation, amorphization temperatures, etc.) of compound materials such as solid solutions of Lanthanides orthophosphates.
  Methods: regression using KRR (Gussian, Laplacian, polynomial) and knowledge extraction through feature sparsification using LASSO + \ell_0.

Work in progress

Ongoing work at FZJ related to ML&DL

Prediction of behavioral scores from MRI markers of brain functional connectivity

Completed projects

tbd

[Home]

last change: 18.4.2019 sw