Week 8 - Generating new data

less than 1 minute read

Meeting summary

  • We have observed that the results of networks trained with modeled data are not as good as they should be.
  • After analyzing the structure of the networks we realize that the main problem is that have not been used the most suitable layers and activation functions for the problem.
  • For this reason we decided to continue testing with different structures that allow us to improve the results.
  • On the other hand, we prioritize the objective of creating new data with different dynamics that allow us to advance in the roadmap.

To Do

The tasks proposed for this week are

  • Generate the datasets with parabolic dynamic.
  • Generate the datasets with sinusoidal dynamic.
  • Try tu optimize the results in networks with modeled data.

Parabolic and sinusoidal frames

In the following link you can find a description of this new type of images.