The first HBP Young Researchers Event took place on April 12 in Budapest. The programme focused on Simulations on different scales of space and time.
The Young Researchers Event took place on April 12 in Budapest, it gathered around 80 participants from 16 countries in Europe. The programme focused on Simulations on different scales of space and time and was open to all young scientists of the HBP community as well as external participants. The goal of this one-day community building event was to train a new generation of scientists to use and collaborate with each other using simulation tools. The programme was developed by a Steering Committee composed of young researchers from various HBP Subprojects, advised by senior HBP researchers and the Education team in Innsbruck. It included compelling sessions such as a keynote lecture, demos and community building.
Download the final programme.
For more information please contact firstname.lastname@example.org
Understanding the higher brain functions is one of the biggest challenges for 21st century science. Brain simulation is one of the approaches to better understand these functions at different scales of space and time. This one-day event provides an overview of elements for simulation both from soft- and hardware perspectives. The programme comprises of several sessions where participants get ample opportunities to exchange information and knowledge about these topics.
The scientific programme is starting with a keynote lecture that will provide a broad overview on simulation. This talk will be followed by a plenary session 'Simulation on different (time) scales', where several facilities developed for simulation of neural networks are presented on an introductory level.
Use the afternoon sessions 'Simulation in use' as an opportunity for extensive demonstrations of different simulation tools. The sessions are taking place in parallel and will be given twice.
And last but not least: If you want to show us your research, use the community building session to present a poster and/or small live demos. An abstract must be submitted for presentation.
Download the final programme.
Requirements for a multi-scale simulation of the transition from deep-sleep to awakeness
by Pier Stanislao Paolucci, Italian National Institute for Nuclear Physics
We will present the (preliminary) requirements for large-scale simulations of Slow Wave Activity and of the transition to awakeness of single and multiple (thalamo-)cortical areas that will be developed by the WaveScalES experiment (WP3.2 in HBP SP3 - Systems and Cognitive Neuroscience). Simulations will be performed using networks of point-like spiking neurons, assembled in two-dimensional columnar-like grids and projecting a few thousands of synapses per neuron. In a first hypothesis, intra-areal inter-columnar connection probabilities will decrease with the distance according to layer specific decay laws (e.g. exponential or Gaussian), while axonal delays will depend on connection length. Inter-areal connection will be projected using (low-complexity) connectivity matrixes. Simulations will initially rely on the proprietary Distributed Plastic Spiking Neural Network (DPSNN) simulation engine developed by INFN in the framework of the EURETILE and CORTICONIC FET Projects. DPSNN demonstrated its efficiency on simulations of several tens of billion of synapses, projected by two-dimensional grids of neural columns, distributed on a thousand of hardware cores and MPI processes. The inclusion of WaveScalES simulations as use-cases of the standard HBP simulation platforms is among our goals. Indeed, this talk aims also to provide a preliminary list of WaveScalES requirements for HBP standard platforms.
MIIND: A Simulator for Population Density Techniques
by Yi Ming Lai, University of Leeds
Population density techniques form a mesoscopic bridge between individual spiking neurons and neural mass models. We start with a large group of spiking neurons; instead of simulating these directly, we formulate equations describing how changes in the state of individual neurons translate to changes in the population state. This allows us to simulate large populations without having to process massive amounts of spiking data, and construct large networks of them. As these populations have been rigorously derived from individual spiking models, these methods also allow us to maintain a level of biological realism, which may be absent in some top-down phenomenological rate-based models.
Simulation of spiking neural networks with Brian 2
by Marcel Stimberg, Inserm
Brian 2 is a fundamental rewrite of the popular Brian simulator for spiking neural networks. It is written in the Python programming language and focuses on simplicity and flexibility: neuronal and synaptic models can be described using mathematical formulae and with the use of physical units. Even though Brian 2 benefits from the ease of use and the flexibility of the Python programming language, its performance is not limited by the speed of Python: at the core of the simulation machinery Brian 2 makes use of fully automated runtime code generation, allowing the same model to be run in the Python interpreter or in compiled C++ code. This presentation will give an introduction to the "Brian way" of describing models and demonstrate the various options to generate efficient code.
Simulating Ion channels, Neurons and Synaptic connections
by Srikanth Ramaswamy
The Blue Brain Project has established a unifying data-driven process for the digital reconstruction of a prototypical neocortical microcircuit. The process unifies a vast body of anatomical and physiological data on ion channel kinetics and distributions, neuron morphologies and electrical types, synaptic kinetics and dynamics to yieldan in silico reconstruction of the cellular and synaptic organization of juvenile rat somatosensory cortex. We present a draft anatomical and physiological map of a prototypical neocortical microcircuit. The map represents the first comprehensive integration of available data and knowledge in a quantitative in silico reconstruction of a part of the brain. To achieve this, a novel predictive strategy was developed using sparse data on the cellular and synaptic organization of the somatosensory cortex of a two-week old rat. The microcircuit is 0.29 mm3 in volume and contains about 31,000 neurons belonging to 55 morphological neuron types and 207 morpho-electrical sub-types distributed across 6 layers.
The resulting reconstruction is broadly consistent with current knowledge about the neocortical microcircuit and provides an array of predictions on its structure and function across neuronal, synaptic, and circuit levels. The models in the reconstruction are available as a public resource for collaborative and iterative refinement, and in silico neuroscience.
by Philipp Weidel, FZ Juelich
NEST (The Neural Simulation Tool, www.nest-simulator.org, github.com/nest/nest-simulator) is an open-source simulator for spiking neural network models that focuses on the dynamics of large structured networks rather than on the exact morphology of individual neurons. The development of NEST is coordinated by the NEST initiative (www.nest-initative.org). NEST is ideal for networks of spiking neurons of any size, for example models of information processing e.g. in the visual or auditory cortex of mammals, models of network activity dynamics, e.g. laminar cortical networks or balanced random networks and models of learning and plasticity. In this 120-minute tutorial, a general introduction to the NEST simulator will be given and demonstrations on how to create and execute NEST models using the PyNEST module in Python will be provided, covering basic and more advanced examples of usage.
Introduction to Heidelberg Neuromorphic Hardware
by Eric Müller/Vitali Karasenko, University of Heidelberg
Over time, several generations of neuromorphic hardware have been developed in Heidelberg. Since the HBP platform release the neuroscience community has gained convenient access to that hardware. We present the different variations of the Heidelberg Neuromorphic Hardware and showcase their usage using the HBP platform
Using the Neuron Simulator
by Werner van Geit, EPFL
In this demo we will give a short introduction on how to use the Python interface to the Neuron simulator. We will show the Neocortical Microcircuit Collaboration Portal, how it can be used to download single cell models that were used in an in silico reconstruction of a neocortical microcircuit, and show how to run these models on desktop computers.
We also introduce a new software tool called BluePyOpt for data-driven model parameter optimisation that wraps and standardises several existing open-source tools.
The SpiNNaker Software Stack
by Alan Stokes, University of Manchester
In this demo, I will show how to power our SpiNNaker software stack, for both PyNN based neural applications and general applications through the use of a Database application. I will also strive to support participants in 1: installing the SpiNNaker software stack, 2: running their own applications on the SpiNNaker machines provided 3: support on how to use the HBP portal. 4: how to use the SpiNNaker software stack in offline mode.
I will also leave myself open to discuss specific functionality provided from the SpiNNaker software stack as requested by the participants during the demo. These can include anything ranging from 1. External device support (such as retinas and motors) 2. live injection / retrieval of events from a executing application, 3. Experiences as a Young scientist 4. whatever comes to mind from the participants.