C10 RNN

Recurrent Neural Networks¶

Often data arise as sequences:

• Documents are sequence of words, and their relative positions have meaning
• Time-series such as weather data or financial indices.
• Recorded speech or music.
• Handwriting, such as doctor's notes.

The feature for each observation is a sequence of vectors $X={X_1,X_2,...,X_L}$

• The target Y is often a single variable such as Sentiment, or a one-hot vector for multiclass.
• Y can also be a sequence, such as the same document in a different language.

• The same wights $W$, $U$ and $B$ are used at each step in the sequence - hence the term recurrent.

Example (IMDB Reviews)¶

Time Series Forecasting¶

Traditional autoregression forecaster¶

When to use Deep Learning¶