Home

Welcome to the introdl wiki!

Questions:

1. What type of Deep Learning model are you considering for the data?
   • We are starting with a Multi-layer Perceptron (see demo/timeseries_rolling.ipynb) and can add more complex models depending on what people's interests are

Follow-up

For those who went through the setup steps on 11/12/16:

• Download tensorflow models here: https://drive.google.com/open?id=0BwfL0l7N-kliNzhuNWozUnQxeFU
• Create a folder named 'results' in introdl/demo/styletransfer

Other resources:

• Demos
• Theory
• Inspiration
• Deep Learning Papers
• Connection with Physics

NASA Datasets Suggestion

Deep learning can be applied to different types of data. Check out deep learning papers and Jason Toy's Creative AI list for ideas on how to apply deep learning to the following NASA Datasets

• Images
  • https://api.nasa.gov/api.html#EPIC
  • https://api.nasa.gov/api.html#MarsPhotos
• Documents
  • https://api.nasa.gov/api.html#patents
• Sounds
  • https://api.nasa.gov/api.html#sounds
• Geometrical data
  • http://hangar.openvsp.org/
• Time series
  • https://search.earthdata.nasa.gov/search/collection-details?p=C179003653-ORNL_DAAC&fst0=Atmosphere&fsm0=Clouds&fs10=Cloud+Amount%2FFrequency

Much more at data.nasa.gov