Advanced search options

Advanced Search Options 🞨

Browse by author name (“Author name starts with…”).

Find ETDs with:

in
/  
in
/  
in
/  
in

Written in Published in Earliest date Latest date

Sorted by

Results per page:

You searched for subject:(synaptic topology). One record found.

Search Limiters

Last 2 Years | English Only

No search limiters apply to these results.

▼ Search Limiters


Penn State University

1. Miller, Aaron James. Characterization of Emergent Synaptic Topologies in Noisy Neural Networks.

Degree: PhD, Physics, 2012, Penn State University

Learned behaviors are one of the key contributors to an animal's ultimate survival. It is widely believed that the brain's microcircuitry undergoes structural changes when a new behavior is learned. In particular, motor learning, during which an animal learns a sequence of muscular movements, often requires precisely-timed coordination between muscles and becomes very natural once ingrained. Experiments show that neurons in the motor cortex exhibit precisely-timed spike activity when performing a learned motor behavior, and constituent stereotypical elements of the behavior can last several hundred milliseconds. The subject of this manuscript concerns how organized synaptic structures that produce stereotypical spike sequences emerge from random, dynamical networks. After a brief introduction in Chapter 1, we begin Chapter 2 by introducing a spike-timing-dependent plasticity (STDP) rule that defines how the activity of the network drives changes in network topology. The rule is then applied to idealized networks of leaky integrate-and-fire neurons (LIF). These neurons are not subjected to the variability that typically characterize neurons \emph{in vivo}. In noiseless networks, synapses develop closed loops of strong connectivity that reproduce stereotypical, precisely-timed spike patterns from an initially random network. We demonstrate the characteristics of the asymptotic synaptic configuration are dependent on the statistics of the initial random network. The spike timings of the neurons simulated in Chapter 2 are generated exactly by a computationally economical, nonlinear mapping which is extended to LIF neurons injected with fluctuating current in Chapter 3. Development of an economical mapping that incorporates noise provides a practical solution to the long simulation times required to produce asymptotic synaptic topologies in networks with STDP in the presence of realistic neuronal variability. The mapping relies on generating numerical solutions to the dynamics of a LIF neuron subjected to Gaussian white noise (GWN). The system reduces to the Ornstein-Uhlenbeck first passage time problem, the solution of which we build into the mapping method of Chapter 2. We demonstrate that simulations using the stochastic mapping have reduced computation time compared to traditional Runge-Kutta methods by more than a factor of 150. In Chapter 4, we use the stochastic mapping to study the dynamics of emerging synaptic topologies in noisy networks. With the addition of membrane noise, networks with dynamical synapses can admit states in which the distribution of the synaptic weights is static under spontaneous activity, but the random connectivity between neurons is dynamical. The widely cited problem of instabilities in networks with STDP is avoided with the implementation of a synaptic decay and an activation threshold on each synapse. When such networks are presented with stimulus modeled by a focused excitatory current, chain-like networks can emerge with the addition of an axon-remodeling plasticity rule, a…

Subjects/Keywords: learning; STDP; neural network; synaptic topology; synfire chain; event-driven simulation; emergent phenomena; complex network

Record DetailsSimilar RecordsGoogle PlusoneFacebookTwitterCiteULikeMendeleyreddit

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6th Edition):

Miller, A. J. (2012). Characterization of Emergent Synaptic Topologies in Noisy Neural Networks. (Doctoral Dissertation). Penn State University. Retrieved from https://etda.libraries.psu.edu/catalog/14822

Chicago Manual of Style (16th Edition):

Miller, Aaron James. “Characterization of Emergent Synaptic Topologies in Noisy Neural Networks.” 2012. Doctoral Dissertation, Penn State University. Accessed December 14, 2019. https://etda.libraries.psu.edu/catalog/14822.

MLA Handbook (7th Edition):

Miller, Aaron James. “Characterization of Emergent Synaptic Topologies in Noisy Neural Networks.” 2012. Web. 14 Dec 2019.

Vancouver:

Miller AJ. Characterization of Emergent Synaptic Topologies in Noisy Neural Networks. [Internet] [Doctoral dissertation]. Penn State University; 2012. [cited 2019 Dec 14]. Available from: https://etda.libraries.psu.edu/catalog/14822.

Council of Science Editors:

Miller AJ. Characterization of Emergent Synaptic Topologies in Noisy Neural Networks. [Doctoral Dissertation]. Penn State University; 2012. Available from: https://etda.libraries.psu.edu/catalog/14822

.