Computational Neuroscience - Research Assistant post
Nigel Crook (ntcrook@brookes.ac.uk)
Tue, 13 May 1997 09:36:44 +0100 (BST)
Research Assistant Post in Computational Neuroscience
Computation in biological and artificial neural networks
The aim of this project is to investigate the computational properties of
biological neurons with a view to applying the results of this investigation
to questions of computation and learning in artificial neural networks (ANNs).
The aim is to apply findings about information processing and storage in
biological neural networks (BNN) to improve the performance of ANNs. A
secondary aim is to contribute to general knowledge and understanding of
information processing in BNNs.
Current ANNs are based on highly simplified models of biological systems, with
synaptic strengths represented as real numbers, the integration of presynaptic
inputs as simple linear summation, and adaptation and learning modelled by
non-biological processes such as backpropagation. Neurophysiological
research, on the other hand, indicates that true biological neurons display
very complex computational and adaptive properties; the dendritic trees of
pyramidal neurons, for example, appear able to extract detailed information
from their neural environments through the spatio-temporal ordering of their
inputs. This information, in turn, may determine the initiation of action
potentials at the axon hillock on its own, or modulate the capacity of basal
synaptic stimuli to initiate action potentials. Additionally, dendritic spines
seem to play a vital role in learning and information processing in cortical
and hippocampal neurons. We have carried out preliminary simulations of simple
dendritic trees using the GENESIS and the NEURON biological simulators and
believe that this work should be continued on a more systematic basis by a
research student.
The aim of this continuing research would be to clarify some of the non-linear
behaviour that dendritic trees implement and to carry these functions over
into new algorithms for information processing and learning in ANNs. We
believe that this could lead to the development of new concepts, models,
topologies and algorithms.
Qualifications
Good honours degree (first or second class) or equivalent in computing or a
cognate discipline is essential.
Application forms and further details of these posts can be obtained from:
Research Admissions,
School of Computing and Mathematical Scineces,
Oxford Brookes University,
Headington,
Oxford OX3 0BP
Tel: +44 (0)1865 483652
Fax: +44 (0)1865 483666
CLOSING DATE: 16 May 1997
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Nigel Crook, Senior Lecturer in Knowledge Engineering __ _____ _____ _____
Intelligent Systems Research Group / / /____/ /___ / /____/
School of Computing & Mathematical Sciences / / //___ //__// //___
Oxford Brookes University, Oxford OX3 0BP, UK / / /___ / / __/ ///_ /
Tel: +44 [0]1865 48371 / / ____// // \\ //__//
Fax: +44 [0]1865 483666 /_/ /____/ // \\/____/
Email: ntcrook@brookes.ac.uk
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
===========================================================================