- Poster presentation
- Open Access
Self-tuning spike-timing dependent plasticity curves to simplify models and improve learning
© Richert et al; licensee BioMed Central Ltd. 2013
- Published: 8 July 2013
- Synaptic Weight
- Neural Model
- Neural Dynamic
- Regular Spike
- Dependent Plasticity
Plasticity for feed-forward excitation ought to optimally assign credit to which synapses cause a postsynaptic cell to spike. It is common to use a double-exponential fit of the LTP and LTD curves ; however, exponential curves are not always optimal and are prone to some pathologies. For example, if there are repeated patterns in the input spikes, learning will degenerate to only the selection of the earliest spikes . Often the parameters for STDP are hand tuned for particular problems and networks.
Measuring of the cross-correlogram offline can provide useful insight as to what the optimal STDP curve should be. We propose an adaptive STDP curve that is derived online from the cross-correlogram, and will discuss its relationship to biology. This dynamic plasticity automatically incorporates an estimate of the dendritic and neuronal integration/processing time in order for a presynaptic input to cause a postsynaptic spike. This plasticity results in faster learning and greater diversity in a model of V1 simple cells. Further, for different neural models and input statistics, different STDP curves will be learned and yet still result in good V1 receptive fields. Because the STDP curve is adaptive to the statistics for each cell, it can be different for each cell in the same population. The model requires only a few meta parameters, which are intuitive, and learning is stable over a large range. Most importantly, instead of having to fiddle with parameters, this synapse model is self-tuning.
- Bi G, Poo M: Synaptic Modifications in Cultured Hippocampal Neurons: Dependence on Spike Timing, Synaptic Strength, and Postsynaptic Cell Type. J Neurosci. 1998, 77: 551-555.Google Scholar
- Masquelier T, Guyonneau R, Thorpe SJ: Spike timing dependent plasticity finds the start of repeating patterns in continuous spike trains. PLoS ONE 3. 2008, e1377-Google Scholar
- Izhikevich EM: Simple model of spiking neurons. IEEE Trans On Neural Networks. 2003, 14 (6): 1569-15. 10.1109/TNN.2003.820440.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.