Sequence Models

Sequence Models

1. Notation

   • X<t> - i-th record of the input sequence
   • Y<t> - i-th record of the output sequence
   • Tx - number of records in the input sequence
   • Ty - number of record in the output sequence
   • T_x^(i) - input sequence length for the training example i

2. NLP

   • Representing words
     • using Vocabulary
       • each word is mapped to number

3. RNN

   • Forward propagation through time
   • Backward Propagation through time

4. GRU (Gated Recurrent Unit)

   • C - memory cell
   • C<t> = a<t>
   • GRU Network uses Gates to flag if the new word should update output, so that it allows to keeps track of the relationships between words further away from each other

Images and files in the map