Sequence Model

Recurrent Neural Network

Cons of Standard Network

  • inputs and outputs can be different length in different examples
  • does not share features learned across different positions

Recurrent Neural Network

Notation

  • Forward Propagation
  • Backpropagation Through Time

Examples

  • many-to-many architecture
  • many-to-one
  • one-to-many

Bidirectional Recurrent Neural Network

results matching ""

    No results matching ""