CONNECTIVITY RECONSTRUCTION ALGORITHM

In this work we present a method of connectivity reconstruction, for the case of pulse-coupled oscillators, inspired by neuroscience. For detailes see the publication.

HOW TO UPLOAD:
Select zip/tar file to upload:

EXAMPLE UPLOAD
NODE DYNAMICS VISUALIZED
COMPLETE PROBLEM VISUALIZATION
Upon runing the algorithm, one obtains
- the connectivity matrix
- natural frequencies
- phase response curves (in terms of fourier coefficients)
- standard deviation error
- the score, the bigger the better (inverse ratio of the error vs the error of the uncoupled system)
t φ = ω     between spikes
φ φ + ε Z( φ)     upon recieving a spike

where ε is the strength of the connection and Z is the
phase responces curve (can be an arbitrary continuous curve).

In the video we at first see a single oscillator and its
phase, with dynamics as described to the left. It is
embeded in a network of oscillators with different
natural frequencies connected with links of different
weights. The phases and connectivity later fade away to
reveal the observations available to us spike trains only.

BACK