ML Anthology

Relating Real-Time Backpropagation and Backpropagation-Through-Time: An Application of Flow Graph Interreciprocity

Françoise Beaufays, Eric A. Wan
NeCo 1994 pp. 296-306
doi:10.1162/NECO.1994.6.2.296 /neco/1994/beaufays1994neco-relating/

Abstract

We show that signal flow graph theory provides a simple way to relate two popular algorithms used for adapting dynamic neural networks, real-time backpropagation and backpropagation-through-time. Starting with the flow graph for real-time backpropagation, we use a simple transposition to produce a second graph. The new graph is shown to be interreciprocal with the original and to correspond to the backpropagation-through-time algorithm. Interreciprocity provides a theoretical argument to verify that both flow graphs implement the same overall weight update.

NeCo Semantic Scholar

Cite

Text

Beaufays and Wan. "Relating Real-Time Backpropagation and Backpropagation-Through-Time: An Application of Flow Graph Interreciprocity." Neural Computation, 1994. doi:10.1162/NECO.1994.6.2.296

Markdown

[Beaufays and Wan. "Relating Real-Time Backpropagation and Backpropagation-Through-Time: An Application of Flow Graph Interreciprocity." Neural Computation, 1994.](https://mlanthology.org/neco/1994/beaufays1994neco-relating/) doi:10.1162/NECO.1994.6.2.296

BibTeX

@article{beaufays1994neco-relating,
  title     = {{Relating Real-Time Backpropagation and Backpropagation-Through-Time: An Application of Flow Graph Interreciprocity}},
  author    = {Beaufays, Françoise and Wan, Eric A.},
  journal   = {Neural Computation},
  year      = {1994},
  pages     = {296-306},
  doi       = {10.1162/NECO.1994.6.2.296},
  volume    = {6},
  url       = {https://mlanthology.org/neco/1994/beaufays1994neco-relating/}
}