Figure - PMC

Skip to main content

An official website of the United States government

Here's how you know

Here's how you know

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

View full-text article in PMC

. 2011 Oct 7;6(10):e25339. doi: 10.1371/journal.pone.0025339

Search in PMC
Search in PubMed
View in NLM Catalog
Add to search

Gilson, Fukai.

This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.

PMC Copyright notice

A: Schematic representation of the network with plastic recurrent connections (thick arrows) and fixed input connections (thin arrows). The network neurons (top blue filled circles) are excited by one pool of 2900 uncorrelated inputs (bottom open circles), and one pool of 100 inputs (bottom red filled circles) whose spike trains may be correlated. B: Time-averaged distributions of the recurrent weights over the learning epoch. Comparison of log-STDP (solid curves) and mlt-STDP (dashed curves) when the small group (red filled circles in A) is uncorrelated (purple) and correlated (red). C: Survival of strong synapses (top 20%) of the distribution over time. The color coding is similar to B. As in Fig. 5C, checks are performed every 5 s and the y-axis indicates the number of surviving synapses from s until the time on the x-axis, cf. (8). D,E,F: Similar curves to B with different starting times . Comparison of D log-STDP with no correlation; E log-STDP with correlations; and F mlt-STDP with correlations. G: Ratio of presence in the top 20% at each check (every 5 s between 200 and 395 s) for the initially strongest at s, cf. (9). Comparison between log-STDP (solid curve) and mlt-STDP (dashed curve) for correlated inputs. The weight indices (x-axis) are sorted. The two horizontal dotted lines indicate 20% and 40%, respectively. H: Distributions of the presence ratio corresponding to G, with a log-scaled y-axis.

Inline graphic — A: Schematic representation of the network with plastic recurrent connections (thick arrows) and fixed input connections (thin arrows). The network neurons (top blue filled circles) are excited by one pool of 2900 uncorrelated inputs (bottom open circles), and one pool of 100 inputs (bottom red filled circles) whose spike trains may be correlated. B: Time-averaged distributions of the recurrent weights over the learning epoch. Comparison of log-STDP (solid curves) and mlt-STDP (dashed curves) when the small group (red filled circles in A) is uncorrelated (purple) and correlated (red). C: Survival of strong synapses (top 20%) of the distribution over time. The color coding is similar to B. As in Fig. 5C, checks are performed every 5 s and the y-axis indicates the number of surviving synapses from s until the time on the x-axis, cf. (8). D,E,F: Similar curves to B with different starting times . Comparison of D log-STDP with no correlation; E log-STDP with correlations; and F mlt-STDP with correlations. G: Ratio of presence in the top 20% at each check (every 5 s between 200 and 395 s) for the initially strongest at s, cf. (9). Comparison between log-STDP (solid curve) and mlt-STDP (dashed curve) for correlated inputs. The weight indices (x-axis) are sorted. The two horizontal dotted lines indicate 20% and 40%, respectively. H: Distributions of the presence ratio corresponding to G, with a log-scaled y-axis.