Week 10 - Training LSTM for modeled samples

less than 1 minute read

Proposed net

I have defined an LSTM network to address prediction with modeled images and recurrent networks. The defined network is as follows:

LSTM Net

Linear dataset results

  • 1 DOF, 8000 training samples, 1000 test samples
LSTM-Linear 1 DOF
  • 2 DOF, 8000 training samples, 1000 test samples
LSTM-Linear 2 DOF

Parabolic dataset results

  • 1 DOF, 8000 training samples, 1000 test samples
LSTM-Parabolic 1 DOF
  • 2 DOF, 8000 training samples, 1000 test samples
LSTM-Parabolic 2 DOF
  • 3 DOF, 80000 training samples, 10000 test samples
LSTM-Parabolic 3 DOF

Sinusoidal dataset results

  • 1 DOF, 80000 training samples, 10000 test samples
LSTM-Sinusoidal 1 DOF
  • 2 DOF, 80000 training samples, 10000 test samples
LSTM-Sinusoidal 2 DOF
  • 3 DOF, 80000 training samples, 10000 test samples
LSTM-Sinusoidal 3 DOF
  • 4 DOF, 80000 training samples, 10000 test samples
LSTM-Sinusoidal 4 DOF

Conclusions

The results improve respect to the non-recurrent structure (MLP). However there is still a prediction limit, this time at 4 DOF sinusoidal dynamic. This shows that recurrence has a great contribution in prediction problems where temporal correlations are important.