Water Tank
  • Home
  • Notebooks
  • API Reference
  1. Learning rules
  2. RLS
  • API reference
  • Layers
    • RecurrentLayer
    • LinearReadout
    • StaticInput
    • TimeSeriesInput
  • Projections
    • connect
  • Learning rules
    • DeltaLearningRule
    • RLS
  • Random distributions
    • Const
    • Uniform
    • Normal
    • Bernouilli
  • Utilities
    • Recorder

On this page

  • RLS
    • Parameters
    • Methods
      • train

RLS

rules.RLS.RLS(self, projection, delta=1e-06)

Recursive least-squares (RLS) learning rule for FORCE learning.

Parameters

Name Type Description Default
projection Projection projection on which to apply the learning rule. required
delta float initial diagonal value of the correlation matrix. 1e-06

Methods

Name Description
train Applies one step of the RLS learning rule.

train

rules.RLS.RLS.train(error)

Applies one step of the RLS learning rule.

Parameters

Name Type Description Default
error np.ndarray error vector at the post-synaptic level. required
DeltaLearningRule
Const
 

Copyright Julien Vitay