An event-driven approach to biologically realistic simulation of neural aggregates
An event-driven approach to biologically realistic simulation of neural aggregates
This thesis explores an emerging alternative to biophysical modelling which exploits the spike-based nature of inter-neuronal communication to replace the continuous simulation framework by a computationally more efficient event-driven technique.
A hierarchical finite state automation neuron model suitable for message-based event-driven simulation (the MBED model) is described and discussed. It encapsulates various aspects of neuronal biophysics: synaptic/axonal latency, finite synapse activation duration, single spike and bursting behaviour, pace making, inhibition driven burst truncation and others.
The message-based event-driven simulator is designed to deliver efficient simulation of large aggregates of MBED neurons, incorporating a customized event queue management algorithm and a strategy for memory-efficient storage of synaptic parameter sets.
Two biological neural systems are tackled utilizing the MBED framework; the locomotory neural circuit of the nematode C. elegans and the mammalian olfactory cortex. The MBED model of the C. elegans locomotory system replicates experimental observations of normal, mutant and laser ablated animals and provides a quantitative description of a rich set of locomotory behaviours. Video recordings of active C. elegans behaviour, an automated image analysis system and a mechanical body model were developed to complement the neuronal simulation.
To further assess the validity of the MBED framework in biological simulations of neuronal aggregates, a model of the olfactory cortex incorporating 105 neurons of three cortical cell classes was developed. The model consistently replicates results obtained experimentally and with the less efficient compartmental technique, while retaining the computational efficiency inherent to event-driven simulation. The typical speed differential between the two techniques is a factor in the range 10-100. The response of the model to shock and random stimuli of various intensities is studied and shown to be in good agreement with previous results.
University of Southampton
Claverol, Enric T
41cfd3d7-ec06-40ae-a607-dc744dd72e84
2000
Claverol, Enric T
41cfd3d7-ec06-40ae-a607-dc744dd72e84
Claverol, Enric T
(2000)
An event-driven approach to biologically realistic simulation of neural aggregates.
University of Southampton, Doctoral Thesis.
Record type:
Thesis
(Doctoral)
Abstract
This thesis explores an emerging alternative to biophysical modelling which exploits the spike-based nature of inter-neuronal communication to replace the continuous simulation framework by a computationally more efficient event-driven technique.
A hierarchical finite state automation neuron model suitable for message-based event-driven simulation (the MBED model) is described and discussed. It encapsulates various aspects of neuronal biophysics: synaptic/axonal latency, finite synapse activation duration, single spike and bursting behaviour, pace making, inhibition driven burst truncation and others.
The message-based event-driven simulator is designed to deliver efficient simulation of large aggregates of MBED neurons, incorporating a customized event queue management algorithm and a strategy for memory-efficient storage of synaptic parameter sets.
Two biological neural systems are tackled utilizing the MBED framework; the locomotory neural circuit of the nematode C. elegans and the mammalian olfactory cortex. The MBED model of the C. elegans locomotory system replicates experimental observations of normal, mutant and laser ablated animals and provides a quantitative description of a rich set of locomotory behaviours. Video recordings of active C. elegans behaviour, an automated image analysis system and a mechanical body model were developed to complement the neuronal simulation.
To further assess the validity of the MBED framework in biological simulations of neuronal aggregates, a model of the olfactory cortex incorporating 105 neurons of three cortical cell classes was developed. The model consistently replicates results obtained experimentally and with the less efficient compartmental technique, while retaining the computational efficiency inherent to event-driven simulation. The typical speed differential between the two techniques is a factor in the range 10-100. The response of the model to shock and random stimuli of various intensities is studied and shown to be in good agreement with previous results.
Text
755176.pdf
- Version of Record
More information
Published date: 2000
Identifiers
Local EPrints ID: 464152
URI: http://eprints.soton.ac.uk/id/eprint/464152
PURE UUID: 93acae92-0fc9-4bc4-96b8-cc46aee10b14
Catalogue record
Date deposited: 04 Jul 2022 21:21
Last modified: 16 Mar 2024 19:18
Export record
Contributors
Author:
Enric T Claverol
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics