On the Asymptotic Equivalence Between Differential Hebbian and Temporal Difference Learning Using a Local Third Factor

Abstract

In this theoretical contribution we provide mathematical proof that two of the most important classes of network learning - correlation-based differential Hebbian learning and reward-based temporal difference learning - are asymptotically equivalent when timing the learning with a local modulatory signal. This opens the opportunity to consistently reformulate most of the abstract reinforcement learning framework from a correlation based perspective that is more closely related to the biophysics of neurons.

Cite

Text

Kolodziejski et al. "On the Asymptotic Equivalence Between Differential Hebbian and Temporal Difference Learning Using a Local Third Factor." Neural Information Processing Systems, 2008.

Markdown

[Kolodziejski et al. "On the Asymptotic Equivalence Between Differential Hebbian and Temporal Difference Learning Using a Local Third Factor." Neural Information Processing Systems, 2008.](https://mlanthology.org/neurips/2008/kolodziejski2008neurips-asymptotic/)

BibTeX

@inproceedings{kolodziejski2008neurips-asymptotic,
  title     = {{On the Asymptotic Equivalence Between Differential Hebbian and Temporal Difference Learning Using a Local Third Factor}},
  author    = {Kolodziejski, Christoph and Porr, Bernd and Tamosiunaite, Minija and Wörgötter, Florentin},
  booktitle = {Neural Information Processing Systems},
  year      = {2008},
  pages     = {857-864},
  url       = {https://mlanthology.org/neurips/2008/kolodziejski2008neurips-asymptotic/}
}