Skip to content

Latest commit

 

History

History
9 lines (8 loc) · 484 Bytes

rnn.md

File metadata and controls

9 lines (8 loc) · 484 Bytes

Recurrent Neural Networks (RNN)

Notes from deeplearning.ai's sequence models course

  • RNNs have "memory"
  • Can read inputs one at a time and remember context through hidden layer activations
  • Uni-directional RNN: take info from past to process later input
  • Bi-directional RNN: take info from both past and future
  • Basic RNNs suffer from vanishing gradients