Chapter 16 Quiz: Recurrent Neural Networks

Chapter 15: Recurrent Networks for Sequential Data

Learning objectives

Test your understanding of RNN hidden states, LSTM gates, vanishing gradients, and sequence modeling

This quiz covers the lecture material from Chapter 16.

RNN: Hidden state $h_t = f(W_h h_{t-1} + W_x x_t + b)$ carries information across time steps.
Vanishing Gradient: $W_h$ multiplied at each step causes gradients to shrink exponentially.
LSTM: 3 gates (forget, input, output). Cell state with additive updates preserves gradients.
GRU: 2 gates (update, reset). Simpler than LSTM, similar performance.
Bidirectional: Process sequence forward and backward. Use when full sequence is available.

Goodfellow, I., Bengio, Y., Courville, A. (2016). Deep Learning, ch. 10. MIT Press.
Hochreiter, S., Schmidhuber, J. (1997). Long short-term memory. Neural Comput. 9(8), 1735–1780.
Mousavi, S.M., Beroza, G.C. (2022). Deep-learning seismology. Science 377, eabm4470.