Spike-Based Compared to Rate-Based Hebbian Learning

Abstract

A correlation-based learning rule at the spike level is formulated, mathematically analyzed, and compared to learning in a firing-rate description. A differential equation for the learning dynamics is derived under the assumption that the time scales of learning and spiking can be separated. For a linear Poissonian neuron model which receives time-dependent stochastic input we show that spike correlations on a millisecond time scale play indeed a role. Corre(cid:173) lations between input and output spikes tend to stabilize structure formation, provided that the form of the learning window is in accordance with Hebb's principle. Conditions for an intrinsic nor(cid:173) malization of the average synaptic weight are discussed.

Cite

Text

Kempter et al. "Spike-Based Compared to Rate-Based Hebbian Learning." Neural Information Processing Systems, 1998.

Markdown

[Kempter et al. "Spike-Based Compared to Rate-Based Hebbian Learning." Neural Information Processing Systems, 1998.](https://mlanthology.org/neurips/1998/kempter1998neurips-spikebased/)

BibTeX

@inproceedings{kempter1998neurips-spikebased,
  title     = {{Spike-Based Compared to Rate-Based Hebbian Learning}},
  author    = {Kempter, Richard and Gerstner, Wulfram and van Hemmen, J. Leo},
  booktitle = {Neural Information Processing Systems},
  year      = {1998},
  pages     = {125-131},
  url       = {https://mlanthology.org/neurips/1998/kempter1998neurips-spikebased/}
}