Week 9 - Training MLP for modeled samples

less than 1 minute read

Proposed net

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

MLP Net

Linear dataset results

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

Parabolic dataset results

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

Sinusoidal dataset results

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

Conclusions

Good results are achieved in all dynamics until reaching the sinusoidal with 2 DOF. In this motion the results worsen and the network begins to lose prediction capacity.

You May Also Enjoy