GeneRec is a generalization of the recirculation algorithm, and approximates Almeida-Pineda recurrent backpropagation.[1][2] It is used as part of the Leabra algorithm for error-driven learning.[3]

The symmetric, midpoint version of GeneRec is equivalent to the contrastive Hebbian learning algorithm (CHL).[1]

See also

References

  1. ^ a b O'Reilly, R.C. Biologically Plausible Error-driven Learning using Local Activation Differences: The Generalized Recirculation Algorithm. Neural Computation, 8, 895–938. Abstract PDF
  2. ^ GeneRec description in Computational explorations in cognitive neuroscience: understanding the mind by Randall C. O'Reilly, Yuko Munakata
  3. ^ Leabra overview in Emergent
Categories