Released v2019.3.1.

Implemented training methods

  • Gradient descent
  • Mini-batch gradient descent
  • Barzilai-Borwein method
  • Levenberg-Marquardt method
  • Global particle swarm optimization
  • Training sequence

Supported architectures

  • Feed-forward (FF)
  • Incomplete FF networks
  • Sets of FF networks with shared weights

Applied functionality

  • Simple ODE/PDE solver

Degree of parallelization

  • Sequential implementation