2018 November 12

Location: Coors Tech

Attending:
Andy Antoine Thomas Iga Jihyun Jonah Bane

TODO:

  • Begin compiling VAE summary (Iga, Andy, Thomas)
  • Begin compiling PyTorch vs. Keras materials (Jihyun)

Problem

  • Bane has a ton of hydro flow models and corresponding data.
  • What geophysical data is most appropriate for detecting flow?
  • Does not have access to flow simulator.

Questions

  • This looks like a sensitivity analysis?
  • What variogram was used to generate your models?

Comments

  • Speed up flow simulation for this particular.
  • Explore how NN performs compared to flow inversion (see Joe C. PhD thesis)
  • Train NN with current model+flow.
  • Recurrent NN seems appropriate.
  • You could generate more models for testing if you had the variogram.
  • “Simplify to a point that is easy to generalize from.” -AG

Website changes

  • Projects tab (coming soon)
  • Jupyter notebook on website? (coming soon)

Topics of discussion for next week

  • Touch on VAEs

Upcoming/backup topics of discussion

  • Variational Auto-encoders summary and application (Iga, Andy, and Thomas)
  • Arnab - existing project
  • PyTorch vs. Keras

Comments