Recurrent Neural Networks¶

Used for time series / sequential data
Long sequences

Derivation of gradient updates for simple RNN¶

LSTM¶

Forget Gate¶

After getting the precious h_{t-1} and the input x_t. Decide how much to forget

\[\]

previous

<no title>

next

<no title>