deeplearning

RNN Notations

Motivating Example

If you want a sequence model to automatically tell you where are the peoples names in this sentence. This is a problem called Named-entity recognition

Each word in X can be noted as $x^{<1>},\cdots, x^{<9>}$

The length of sequence X is $T_x=9$

Similary each y can be noted as $y^{<1>},\cdots, y^{<9>}$

The length of sequence Y is $T_y=9$

The length of sequence $X$ of sample $i$ is $T_x^{(i)}$

vocabulary is represented by

$$V=\begin{bmatrix} w_1^{(i)}\\ w_2^{(i)}\\ \vdots\\ w_D^{(i)} \end{bmatrix}$$

One-hot encoding

Unknown word will be represented as