% IMPORTANT: The following is UTF-8 encoded.  This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.

@INPROCEEDINGS{Wischnewski:905258,
      author       = {Wischnewski, Kevin and Eickhoff, Simon and Popovych,
                      Oleksandr},
      title        = {{E}fficient validation of dynamical whole-brain models via
                      mathematical optimization algorithms},
      reportid     = {FZJ-2022-00541},
      year         = {2021},
      abstract     = {Investigating the resting-state brain dynamics involves its
                      simulation via mathematical whole-brainmodels. The quality
                      of the models’ output and its benefit for further studies,
                      however, depend onoptimally selected input parameters, whose
                      detection via systematic parameter space scans
                      becomespractically unfeasible for high-dimensional models.
                      In our work, we thus test and analyze severalalternative
                      approaches to solve the so-called inverse problem that
                      consists of a parameter-dependentmaximization of the
                      models’ goodness-of-fit to empirical data. An exhaustive
                      parameter variationon a dense grid serves as a benchmark to
                      assess the performance of four optimization
                      schemes:Nelder-Mead Algorithm (NMA), Particle Swarm
                      Optimization (PSO), Covariance Matrix AdaptationEvolution
                      Strategy (CMAES) and Bayesian Optimization (BO). To compare
                      the methods, we employa dynamical model of coupled phase
                      oscillators built upon the individual empirical structural
                      connectivities of a cohort of 105 healthy subjects. For each
                      subject, we determine the optimal modelparameters from two-
                      and three-dimensional parameter spaces to maximize the
                      correspondencebetween simulated and empirical functional
                      connectivity. We show that the overall fitting quality of
                      the tested methods can compete with the extensive parameter
                      sweep exploration. There are,however, marked differences in
                      the required computational resources and stability
                      properties ofthe investigated techniques. By considering a
                      trade-off between enhanced global convergence andthe economy
                      of computation time, we propose two approaches, CMAES and
                      BO, as effective andresource-saving alternatives to a
                      high-dimensional parameter search on a dense grid. For the
                      three-dimensional parameter optimization, they generated
                      similar results as the grid search, but withinless than
                      $6\%$ of the required computation time. Our results can
                      contribute to an efficient validationof mathematical models
                      for personalized simulations of brain dynamics.},
      month         = {Oct},
      date          = {2021-10-05},
      organization  = {INM $\&$ IBI Retreat 2021,
                       Forschungszentrum Jülich, Virtual
                       Conference (Germany), 5 Oct 2021 - 6
                       Oct 2021},
      subtyp        = {After Call},
      cin          = {INM-7},
      cid          = {I:(DE-Juel1)INM-7-20090406},
      pnm          = {5232 - Computational Principles (POF4-523) / HBP SGA2 -
                      Human Brain Project Specific Grant Agreement 2 (785907) /
                      HBP SGA3 - Human Brain Project Specific Grant Agreement 3
                      (945539) / VirtualBrainCloud - Personalized Recommendations
                      for Neurodegenerative Disease (826421)},
      pid          = {G:(DE-HGF)POF4-5232 / G:(EU-Grant)785907 /
                      G:(EU-Grant)945539 / G:(EU-Grant)826421},
      typ          = {PUB:(DE-HGF)24},
      url          = {https://juser.fz-juelich.de/record/905258},
}