Training Algorithm

The training code evaluates each input using :func:’lapart.train.lapart_train’

The supervised training algorithm used known inputs and outputs, and free parameters to create templates TA, TB, and L.

train.lapArt_train(xA, xB, rhoA=0.9, rhoB=0.9, beta=1e-06, alpha=1.0, nep=1, memory_folder='', update_templates=True, normalize_data=True)[source]

Train LAPART Algorithm

  • xA – A-Side Input Matrix (float)
  • xB – B-Side Input Matrix (float)
  • rhoA – A-Side free parameter (float)
  • rhoB – B-Side free parameter (float)
  • beta – Learning rate free parameter (float)
  • alpha – Choice Parameter (float)
  • nep – Number of epochs (integer)
  • memory_folder – Folder to store memory (string)
  • update_templates – Command to update or create new templates (boolean)
Return TA:

A-Side template matrix (float)

Return TB:

B-Side template matrix (float)

Return L:

Associator matrix (float)

Return elapsed_time:

Seconds to complete training (float)