HBP concluding event

Title: Macroscopic and topographical anatomy of the human brain

Abstract:

Introduction to human brain macroanatomy in the framework of functional systems: Gyri, sulci and major subcortical nuclei will be demonstrated on fixed human brains and coronal section series. Participants will receive an introduction into the macroscopic and topographical anatomy of the human brain in relation to functional systems. Plastinated preparations of whole brains and of coronal section series will be available, in which they will be able to identify major landmarks such as the central sulcus, different subcortical nuclei or the Gennari stripe.

Organisers and speakers: 

Nicola Palomero-Gallagher (Forschungszentrum Juelich)

Preparations & Equipment:

• no technical equipment required

Title: NEST Desktop

Abstract:
NEST Desktop is a graphical user interface for the NEST Simulator, a well-established simulator in computational neuroscience that celebrates its 25th birthday this year. Like most simulators, NEST is used by scripting language, which requires a user to have basic programming skills. This poses a problem for beginners, as experience shows that they usually have little to no coding experience. NEST Desktop fills this gap and serves as a very intuitive interface to the NEST Simulator and has already been used in official student courses at universities in parts of Europe.
This hands-on session provides an introduction to NEST Desktop and NEST.
NEST Desktop is a web-based graphical user interface (GUI), which allows the exploration of essential concepts in computational neuroscience without the need to learn a programming language. This advances both the quality and speed of teaching in computational neuroscience. To get acquainted with the GUI, we will create and analyze a balanced two-population network.

Organisers and speakers: 

Sebastian Spreizer (University of Trier)
Jens Bruchertseifer (University of Trier)

Preparations & Equipment:

Each user should register for an EBRAINS account in advance. A Notebook/PC with internet connection and an up-to-date web browser (Chrome or Firefox recommended) will be needed.

Title: The siibra toolsuite for working with multilevel brain atlases

Abstract:

This session is an introduction to the interactive online tools and programmatic access to EBRAINS brain atlases.

Organisers and speakers: 

Timo Dickscheid (Forschungszentrum Juelich)
Kimberley Lothmann (Forschungszentrum Juelich)

Preparations & Equipment:

Each user should register for an EBRAINS account in advance. A Notebook/PC with internet connection and an up-to-date web browser (Chrome or Firefox recommended) will be needed.

Title: Modular Science/ Multi scale session

Abstract:

The Modular Science/ Multi scale session will provide an overview and introduction of the recent advances in deeply integrated multi-scale, multi-model co-simulation workflows including integration of Insite streaming processing, monitoring of real-time system states and deployment of NEST Desktop created workloads.

Organisers and speakers: 

Wouter Klijn (Forschungszentrum Juelich)

Preparations & Equipment:

A Notebook/PC with internet connection and an up-to-date web browser (Chrome or Firefox recommended) will be needed. The Vagrant and Virtual Box should be installed.

Title: BrainScaleS neuromorphic computing hands-on session

Abstract:

In the BrainScales hands on session we will introduce the audience into using the BrainScaleS-2 neuromorphic compute systems interactively via the EBRAINS "Lab". Participants will have the possibility to use the system live with several tutorial notebooks using their own  laptop. The tutorial notebooks cover a wide range of usage possibilities from single neuron demonstrations via structured neurons to online learning.

Organisers and speakers: 

Bjoern Kindler (Kirchhoff Institute for Physics, University of Heidelberg)

Preparations & Equipment:

A Notebook/PC with internet connection and an up-to-date web browser (Chrome or Firefox recommended) will be needed.

Title: NeuroCONNECT – a versatile brain MR imaging tool for brain damages and cognitive decline

Abstract:
In this workshop we will introduce NeuroCONNECT as a versatile brain MR imaging tool in research and clinical applications. The hands-on session will allow you to explore various multiple MR image feature results. These include voxel- and surface-based morphometry analyses on anatomical images, tractography on diffusion weighted images (DTI), and functional connectivity approaches on resting-state fMRI data. Extracted imaging features are integrated based on the atlas-defined region of interest (ROI). You may explore various connectome outcomes reconstructed by the Julich atlas from EBRAINS, the HCP MMP 1.0 atlas, and classical atlases like Schaefer, Freesurfer and Harvard-Oxford atlases to be visualized in a unified feature space, called NeuroCONNECT 3D feature viewer. Atlas ontologies form the platform for knowledge-based feature integration under the hood of NeuroCONNECT.

NeuroCONNECT runs from an easy-to-use webportal. The portal can be accessed for free trials within the HBP community. Accounts for the NeuroCONNECT portal can be obtained ahead of the workshop for HBP partners and workshop participants by sending an email to josine.verhaal@labvantage-biomax.com. Please bring a laptop with chrome or safari browser installed.

1.    Understand how the standardized space in NeuroCONNECT works to integrate information from more than one MRI modality
2.    Understand how connectome data can be mapped to more than one atlas
3.    Learn how you can speed up your research by analyzing MR imaging features with metadata and specific measurements in NeuroCONNECT

Organisers and speakers: 

Josine Verhaal (Labvantage - Biomax GmbH)
Sebastian Kopetzky (Labvantage - Biomax GmbH)
Benedikt Weiss (Labvantage - Biomax GmbH)
Josine Verhaal (Labvantage - Biomax GmbH)
Yong Li (Labvantage - Biomax GmbH)

Preparations & equipment:
A Notebook/PC with internet connection and an up-to-date web browser (Chrome or Firefox recommended) will be needed.

Title: Handling EBRAINS data

Abstract:

Introduction for handling data in the EBRAINS Knowledge Graph using the KG Core Python SDK and the QueryBuilder, as well as the Python package fairgraph.

Organisers and speakers: 

Lyuba Zehl (Forschungszentrum Juelich)
Andrew Davison (CNRS)
Oliver Schmid (EBRAINS AISBL)

Preparations & equipment:

A Notebook/PC with internet connection and an up-to-date web browser (Chrome or Firefox recommended) will be needed, Python knowledge is required.

JSC Quantum Computer Tour

JUNIQ - the Jülich UNified Infrastructure for Quantum computing - provides access to state-of-the-art quantum computing devices to science and industry. We will give an overview on activities and how to access quantum computers within JUNIQ, and give an introduction to quantum computing with the focus on quantum annealing.

JSC Supercomputer Tour

The Jülich Supercomputing Centre provides leading-edge supercomputer resources, IT tools, methods, and know-how. It operates one of the most powerful supercomputers in Europe, JUWELS. We will give an overview on our activities and show how we make our infrastructure made available to researchers in Germany and Europe.

INM-1 wetlab tour

During the guided tour you will visit the wetlab, the lab for Porlarized Light imaging and the scanner farm of the Department of Structural and Functional Organization of the Brain (INM-1). 

Prof. Katrin Amunts did a postdoctoral fellowship at the C. & O. Vogt Institute of Brain Research at Duesseldorf University, Germany. In 1999, she set up a new research unit for Brain Mapping at the Research Center Juelich, Germany. In 2004, she became professor for Structural-Functional Brain Mapping, and in 2008 a full professor at the Department of Psychiatry, Psychotherapy and Psychosomatics at the RWTH Aachen University as well as director of the Institute of Neuroscience and Medicine (INM-1) at the Research Center Juelich. Since 2013, she is a full professor for Brain Research, director of the C. and O. Vogt Institute of Brain Research, Heinrich-Heine University Duesseldorf and director of the Institute of Neuroscience and Medicine (INM-1), Research Center Juelich.
Katrin Amunts is a member of the editorial board of Brain Structure and Function. She is member of the German Ethics Council since 2012, and has been elected as vice chair in 2016. Katrin Amunts is the programme speaker of the programme Decoding the Human Brain of the Helmholtz Association, Germany. She is leading Subproject 2 Human Brain Organization of the European Flagship Project The Human Brain Project (HBP). In 2016, she has been elected as Scientific Research Director and Chair of the Science and Infrastructure Board (SIB) of the HBP. Since 2017 Katrin Amunts is co-speaker of the graduate school Max-Planck School of Cognition and since 2018 she is a member of the International Advisory Council Healthy Brains for Healthy Lives, Canada.
In order to better understand the organizational principles of the human brain, she and her team aim to develop a multi-level and multi-scale brain atlas, and use methods of high-performance computing to generate ultra-high resolution human brain models.

Prof. Thomas Lippert is the director of the Institute for Advanced Simulation (IAS) at Forschungszentrum Jülich, Germany. He is the head of the Jülich Supercomputing Centre (JSC), a division of the IAS, and acts as managing director of the John von Neumann Institute for Computing (NIC). He is professor for Modular Supercomputing and Quantum Computing at the Goethe University Frankfurt, Germany. He is chair of the board of directors of the Gauss Centre for Supercomputing e.V. From 2018 to 2020 (June), he was chair of the council of the Partnership for Advanced Computing in Europe (PRACE). Currently, he acts as German representative at the PRACE council. Since 2019, he has been spokesman of the Helmholtz Programme “Engineering Digital Futures: Supercomputing, Data Management and Information”. He has initiated the series of Europe founded DEEP projects. In 2022, he was elected vice chair of the RIAG of the EuroHPC JU.

Prof. Rainer Goebel studied psychology and computer science in Marburg (1983-1988) and completed his PhD in 1994 at the TU Braunschweig, Germany. He received the Heinz Maier Leibnitz Advancement award in cognitive science in 1993, and the Heinz Billing award from the Max Planck society in 1994 for a neural network software package. From 1995-1999 he was a postdoc at the Max Planck Institute for Brain Research in Frankfurt/Main. Since January 2000, he is professor for Cognitive Neuroscience at Maastricht University, Netherlands. He is founding director of the Maastricht Brain Imaging Centre. Since 2014 he is member of the Royal Netherlands Academy of Arts and Sciences and since 2017, member of Leopoldina, the German National Academy of Science. He received an ERC Advanced Investigators Grant (2011 - 2016). He currently leads one of three science work packages of the Human Brain Project (HBP) at the interface of neuroscience, AI and robotics.

Prof. Petra Ritter heads the Brain Simulation Section at the Charité University Medicine Berlin and Berlin Institute of Health. Her research focus is on integrating neuroimaging and computational neuroscience to discover mechanisms of brain function and dysfunction. She serves in the leadership of large EU projects such as Testing and Experimentation Facility Health AI and Robotics (TEF-Health), Virtual Brain Cloud & eBRAIN-Health and is directing EBRAINS Health Data Cloud. Petra Ritter studied medicine at the Charite and in the US. At the Charité she got appointed a Johanna Quandt lifetime Professorship for Brain Simulation in 2017. Since 2017, she is also Director of the Brain Simulation Section at Charité Universitätsmedizin Berlin. She also serves as the Director for International Affairs at the Charité.

Dr. Sebastian Billaudelle is a postdoctoral researcher at the Kirchhoff-Institute for Physics at Heidelberg University, Germany, where he received his Ph.D. degree in physics in 2022. His scientific work spans from the design of mixed-signal neuromorphic circuits and systems to accompanying algorithms and their applications in biology- and machine-learning-inspired settings.

Talk: Flexible silicon for neuron emulation and efficient computation

Neuromorphic computation seeks to cover a highly diverse spectrum – from the emulation of neurobiological dynamics and networks to the exploration of efficient and fast information processing paradigms. The accelerated neuromorphic BrainScaleS-2 system, in particular, attempts to bridge this range and positions itself both as a tool for computational neuroscientists and as a platform for the exploration of efficient spike-based computation. Catering to these very different applications requires flexible and reliable circuits, accompanied by algorithms capable of taking full advantage of the circuit design. This talk briefly introduces some of the circuit design choices and algorithmic approaches. It then demonstrates select applications of BrainScaleS-2 with focus on its silicon neurons. These examples range from the biology-inspired emulation of complex neuron dynamics to a showcase of gradient-based training of the neuromorphic system motivated by machine learning. The latter encompasses a demonstration of energy-efficient and low-latency classification and continues to more complex examples involving memory on longer time scales provided, e.g, by the adaptation mechanism of the silicon neuron.

Jens Bruchertseifer received his Bachelor's degree in computer science from Trier University. In addition to his master studies, he is currently working as a research associate at Trier University within the Human Brain Project on steering and visualization methods for HPC simulations. So far he could gain experience not only in practical, but also in theoretical areas like parameterized algorithmics, which he applied in his Bachelor's thesis as well as in further work beyond the studies.

Talk: NEST Desktop

NEST Desktop is a web-based graphical user interface (GUI) which comprises graphical elements for creating and configuring network models, running simulations in the NEST Simulator, and visualizing and analyzing the results. It allows students to explore important concepts in computational neuroscience without the need to learn a simulator control language beforehand. This lowers the barrier for newcomers to learn computational neuroscience.    This talk gives an introduction to NEST Desktop and displays multiple scenarios of use cases, also with external tools. Since the latest release, NEST Desktop e.g. allows users to investigate plasticity and contains a model manager to work on the characteristics of neuron models.    In order to give students, teachers, and researchers installation-free access to the compute resources, we integrated NEST Desktop into the EBRAINS infrastructure. The same code remains available as a stand-alone version of NEST Desktop for applications in teaching and training and installations at other sites.

Dr. Leonardo Dalla Porta is a postdoctoral researcher in the laboratory of Dr. Maria V. Sanchez-Vives at IDIBAPS, Barcelona, Spain. With a background in physics and computational neuroscience, his research focuses on the intricate relationship between brain states and cortical complexity. He primarily works with electrophysiological data analysis and mathematical simulations to unravel the brain's functional organization and elucidate the mechanisms governing cortical state transitions.

Talk: From ionic channels to global brain dynamics and behaviour: data-driven simulations of cholinergic innervation

There is an intricate relationship between brain states and complexity, stemming from the dynamic interplay of neural activity and the emergence of cognitive phenomena. During slow wave sleep (SWS) we are unconscious and disconnected from the outside world. In contrast, when we awaken, we become self-aware and conscious of the world that surrounds us. These states differ dynamically: SWS is highly synchronous with low complexity, while wakefulness is desynchronized with high complexity. However, what are the mechanisms driving the transition between these very different functional brain states? Our work, which involves efforts made in Working Package 2 (HBP-WP2), took a bottom-up approach. By analyzing data from isolated cortical circuits, we showed the impact of neuromodulator acetylcholine through muscarinic action on the slow wave’s dynamics. Its action resulted in a depolarization of neurons and deeply transforming cortical dynamics nearly approaching a desynchronized state. We replicated these finds in a biophysically detailed network model of a local network. Integrating scales with The Virtual Brain, we used mean-field modeling of Adaptive Exponential neurons together with a connectome including cholinergic-muscarinic innervation. This provided global dynamics to quantify wave propagation and synchronization across different brain states. Furthermore, local stimuli were used to investigate response propagation, recruitment of higher areas, and information processing.

After PhD in computer science focused on HPC and complex networks I shifted my focus fully to the field of computational neuroscience. Currently I'm working on topics centered around brain network modeling and inference, mainly in the context of healthy aging, neurodegenerative diseases, and noninvasive stimulation. As a member of the The Virtual Brain Facility Hub I'm participating in the EBRAINS technical coordination and assist with formulation and execution of use cases building on the TVB EBRAINS services.

Talk: Showcase 1 - the virtual aging brain: structure-function relationship and cognitive decline

The human brain changes during healthy aging with large individual variation in the cognitive decline. These changes are linked with whole-brain reorganisation, specifically in terms of the white-matter fibre tracts (structural connectivity, SC), and functional co-activations (functional connectivity, FC). Here we present a causal framework of virtual twin modelling where the individual SC informs a computational brain network model. Within this framework we can operationalize hypothesised mechanisms of aging individually by modifying specific SC links, and systematically evaluate the impact on the brain dynamics. Our results indicate that the deterioration of the SC is accompanied by increased modulation of the functional brain dynamics. Across the 1000BRAINS dataset, we identified a strong decline in interhemispheric SC connections over age accompanied with decline both in homotopic FC and FCD fluidity. The trends for both these functional features were recovered in simulated resting state data within the virtual model and also for the virtually aged subjects along the individual trajectories. The SC reorganisation might reflect scaffolding of the brain during the ageing process. This effect is weaker for the cognitively well performing subjects, which suggests a process of brain maintenance. The implementation of the Showcase is fully embedded in EBRAINS and freely available, providing a reusable template for researchers aiming to use EBRAINS services in their work.

Archana Golla is a researcher affiliated with the University of Oslo, Norway. Her academic career began with a bachelor’s and master’s degrees in Biotechnology from VIT University, India. In 2021, she received her doctoral degree in neuroscience from NTNU, Norway. Her thesis focused on the impact of early-life stress on behavior and neuronal responses in zebrafish. Archana currently works at the EBRAINS research platform as a scientific data curator/coordinator and also develops metadata tools.

Talk: EBRAINS Data & Knowledge services: SHARE and FIND data

EBRAINS is a digital research infrastructure created by the Human Brain Project and committed to open neuroscience. The Data & Knowledge Services at EBRAINS aim to ensure viable and FAIR data sharing. The services rest on three pillars: curation, Knowledge Graph, and openMINDS metadata framework.  Our expert-run curation service plays a crucial role in enabling the FAIR sharing of neuroscience data, software, and models on the EBRAINS platform. Working closely with researchers, our curators guide them through the process of structuring, describing, assigning a license for use, and annotating the output of their research. This ensures that the research is made more easily discoverable, accessible, interpretable, and re-usable as well as linked to other relevant research.   The Knowledge Graph brings together the curated metadata collected from all entries and creates linking of instances, enabling their reuse. At the core of the EBRAINS Knowledge Graph lies a graph database that connects neuroscientific research across different modalities. This integration is made possible by leveraging the openMINDS metadata framework that provides a standardized set of metadata models, along with libraries of controlled terminologies, brain atlases, and common coordinate spaces.  By building upon these three fundamental components, the Data & Knowledge Services of EBRAINS strive to combine multifaceted neuroscience data, enhance reproducibility and replicability, and foster collaboration. 

Studying Physics at the RWTH Aachen University, Robin Gutzen developed a deep interest in Neuroscience. During his Master’s he joined the Institute for Systems and Computational Neuroscience in Jülich, working under Michael Denker and Sonja Grün on statistical validation methods of network activity. In his recently completed PhD, he investigated the dynamics of neural networks by developing analytical tools to capture and compare their ensemble activity, which he now employs towards new applications.

Talk: Exploring the diversity of cortical wave activity with a unifying workflow approach

Although brain waves have been studied for a long time, the complex spatial dynamics of such waves became observable only with high-resolution measurement technology. Recent studies of cortical wave activity present various propagation patterns, cortical localization, frequency regimes, and potentially functional roles [e.g. Denker et al. 2018,  Davis et al. 2020].  Such heterogeneity warrants analysis approaches that enable the combination and comparison of data and results from different sources, facilitating a cumulative understanding of cortical wave activity. We developed an adaptable and reusable workflow approach to combine various data modalities and analysis methods, by combining existing software tools and standards from the EBRAINS environment.  We showcase how this approach enables large meta-studies, comparing slow wave activity in anesthetized mice across heterogenous data sources [Gutzen et al. 2022] and the calibration and validation of corresponding network models [Capone et al. 2023]. Further, we demonstrate its extension to new applications, data modalities, and wave types. Specifically, we analyze LFP phase waves in the visual cortex of the macaque monkey in the context of visual perception and the coordination of corresponding saccadic eye movements.  In doing so, we illustrate how the adaptable reusability of research software accelerates research, enhances automation, supports collaboration, promotes reproducibility, and enables cross-domain comparisons.

Camilla Krämer obtained her M.Sc. in Neuro-Cognitive Psychology from Ludwig-Maximilians-Universität in Munich in 2017. Since 2019 she is doing her PhD at the Institute for Anatomy I, Heinrich Heine University Düsseldorf, Germany and the Connectivity group of the Institute of Neuroscience and Medicine (INM-1), Research Centre Juelich, Germany. She is focusing her research on the prediction of behaviour from multimodal imaging data. Particularly, working with data of the population-based cohort 1000BRAINS, she investigates whether cognitive performance differences may be predicted from brain network information in older adults.

Talk: Differential predictability of cognitive profiles from brain structure in older males and females

Structural brain imaging parameters may successfully predict cognitive performance in neurodegenerative diseases, but mostly fail to predict cognitive abilities in healthy older adults. One important aspect contributing to this might be sex differences. Behaviorally, older males and females have been found to differ in terms of cognitive profiles, which cannot be captured by examining them as one homogenous group. In the current study, we examined whether the prediction of cognitive performance from brain structure, i.e. region-wise grey matter volume (GMV), would benefit from the investigation of sex-specific cognitive profiles in a large sample of older adults (1000BRAINS; N= 634; age range 55-85 years). Prediction performance was assessed using a machine learning (ML) approach. Targets represented a) a whole-sample cognitive component solution extracted from males and females, and b) sex-specific cognitive components. Results revealed a generally low predictability of cognitive profiles from region-wise GMV. In males, low predictability was observed across both, the whole sample as well as sex-specific cognitive components. In females, however, predictability differences across sex-specific cognitive components were observed, i.e. visual working memory (WM) and executive functions showed higher predictability than fluency and verbal WM. Hence, results accentuated that addressing sex-specific cognitive profiles allowed a more fine-grained investigation of predictability differences, which may not be observable in the prediction of the whole-sample solution. The current findings therewith not only emphasize the need to further investigate the predictive power of each cognitive component, they also emphasize the importance of sex-specific analyses in older adults.

I’m a research fellow at APE lab, INFN (Rome), enrolled in the framework of EBRAINS-Italy project. With a background in statistical physics, I got the PhD at Sapienza (Rome), working with F. Ricci-Tersenghi and G. Parisi on disordered glassy systems. I then moved to Paris for a post-doc, working with A. Walczak and T. Mora at the Ecole Normale Superieure on computation biophysics applied to immunology. My experience in computational neuroscience dates back to 2021, when I joined HBP-EBRAINS.

Talk: Towards an EBRAINS service for brain wave analysis: Cobrawap

Cobrawap (Collaborative Brain Wave Analysis Pipeline) is a FAIR-compliant open-source software cooperatively developed as a HBP-EBRAINS SGA3 UseCase, aiming at standardizing quantitative descriptions of brain wave phenomena. It is written in Python3 and structured as a collection of modular building blocks (that can be added, removed or replaced) arranged along sequential stages, implementing data processing steps and analysis methods, directed by a workflow manager.  Aiming at a wider community diffusion and at user facilitation, recent efforts have been addressed to integrate Cobrawap as an EBRAINS component. Among the main achievements in this regard, we succeeded in deploying Cobrawap on HPC sites belonging to the FENIX-ICEI federation, making it accessible through ssh direct login, from the EBRAINS Collab through UNICORE and eventually from the Workflows Dashboard.  A crucial step toward this result is represented by the development of dedicated CWL workflows, which are dynamically built at runtime through a “meta-approach” parsing user-friendly fully customizable configuration files. The goal is to minimize the time-to-result, letting the user focus on the scientific side without caring of the technology behind the scenes.  Pushing further the portability and the user-friendliness of Cobrawap, we are going to deliver it as both a Docker image and a pip-installable Python package, regularly upgraded on both scientific and technological sides through a dedicated CI/CD pipeline.

Eleni Mathioulaki is a computer engineer at IMSI, Athena RC. She received her M.Eng. with a specialization in Computer Systems/Software from NTUA, Greece in 2020. Her research interests include Machine Learning, Computer Vision, and Data Science. For the past two years she's been part of the Technical Coordination team of HBP, where her main focus is the establishment of automated procedures for the delivery and deployment of EBRAINS software and the standardization of computational workflows.

Talk: HPC deployment of the official EBRAINS software releases

Building and deploying EBRAINS simulation engines and tools in a common environment and making them available to users poses significant challenges due to the diverse range of tools and the increasing complexity of resolving conflicting requirements. To address this challenge, a delivery strategy leveraging Spack, a specialized package manager for scientific software, has been implemented. This approach has led to the development of standardized, versioned, tested "official EBRAINS software releases", that encompass nearly all EBRAINS tools and have been successfully deployed to the Collaboratory Lab environment for over a year.  In order to further enhance the usability of EBRAINS software on High-Performance Computing (HPC) resources and promote reproducibility in computational experiments and workflows, a robust mechanism was needed to make those EBRAINS software releases available on FENIX scalable and interactive computer systems. To achieve this, a dedicated team collaborated closely with HPC administrators, with the goal of understanding the capabilities and constraints of FENIX systems and devising an optimised, streamlined deployment process tailored to the unique requirements of EBRAINS software. This integration empowers researchers to leverage the powerful capabilities of HPC resources for their computational workflows while ensuring the reproducibility and portability of their experiments.

Michiel Alexander van der Vlag specializes in High Performance Compute solutions for neuroscientific simulations at the simlab for lifesciences at the Juelich Supercomputing Centre. Knowledge of electrical engineering, computer science and philosophy enables him to address problems interdisciplinarily. His expertise lies in knowledge of performant hardware (GPU, CPU, fpga) and knows many programming languages, enabling the enrichment of computational neuroscience with the power of supercomputers.

Talk: Leveraging learning to learn for intelligent parameter space exploration on HPC using GPUs for The Virtual Brain

The Virtual Brain (TVB; Sanz Leon et al. 2014) is a neuroinformatics platform for full brain network simulations using biologically realistic connectivity. In order to abstract implementation complexity of TVB’s whole brain network models, the model generator RateML (Vlag et al. 2022) can be utilized. One of its outputs is a model and simulator object for the Compute Unified Device Architecture (CUDA) platform, enabling parameter exploration by making use of the parallel architecture of the GPU. In order to find certain brain regimes for this model intelligently, a scalable hyper parameter optimization framework called L2L, implementing the concept of learning to learn, can be instantiated to do parameter space exploration in an automated and parallel fashion (Yegenoglu et al. 2022). L2L is agnostic to the process to be optimized, called optimizee, its optimizers use population based decision algorithms, and simulations can run embarrassingly parallel. A simulation which has been used in a study for scale-integrated understanding of conscious and unconscious brain states and their mechanisms (Goldman et al. 2021), is used as a demonstrator to efficiently find desired compound parametrizations, using the structure to function correlation as fitness. This workflow (which will be presented), from model generation to intelligent parameter exploration, which is an integral part of WP1 showcase 1, makes the computational infrastructure on the supercomputer accessible to scientists.