What is it about?
This paper proposes an STDP-like algorithm to decode polychronous neuronal groups (PNGs) -- a type of cell assembly. The readout neuron equipped with STDP can eventually develop weights and delays selective to the spatiotemporal pattern in the PNGs from noisy background spikes. The readout neurons can form multiple layers where hierarchical PNGs can be decoded.
Featured Image
Why is it important?
The existing algorithms are not biologically plausible, i.e. by artificially scanning through the network structure. The rich spatiotemporal pattern permitted by polychronous neuronal groups actually leads to a huge search space for such scanning-based algorithms. The algorithm presented in this paper is innovative and biologically plausible, since it uses STDP which is a local updating algorithm without the need for any clever scanning strategies.
Perspectives
The concept of cell assembly is not new, as proposed by Hebb in the 1950s'. Currently, polychronous neuronal group (PNG) is still a hypothetical form of implementing cell assembly. If PNG really exists in biological neuronal systems, hopefully be verified in biological experiments in the future due to technology advance, the algorithm described in this paper could be a candidate with high priority to biologically decode PNGs.
Haoqi Sun
Read the Original
This page is a summary of: Learning Polychronous Neuronal Groups Using Joint Weight-Delay Spike-Timing-Dependent Plasticity, Neural Computation, October 2016, The MIT Press,
DOI: 10.1162/neco_a_00879.
You can read the full text:
Contributors
The following have contributed to this page
