Recurrent Neural Networks

Introduction

RNNs process sequential data by maintaining internal state across timesteps.

Simple RNN

from tensorflow.keras import layers

model = keras.Sequential([
    layers.Embedding(10000, 64, input_length=100),
    layers.SimpleRNN(64),
    layers.Dense(1, activation="sigmoid")
])

LSTM

model = keras.Sequential([
    layers.Embedding(10000, 64, input_length=100),
    layers.LSTM(64, return_sequences=True),
    layers.LSTM(32),
    layers.Dense(1, activation="sigmoid")
])

GRU

model = keras.Sequential([
    layers.Embedding(10000, 64, input_length=100),
    layers.GRU(64),
    layers.Dense(1, activation="sigmoid")
])

Practice Problems

1. Build RNN for text classification
2. Use LSTM for sequence modeling
3. Compare RNN architectures
4. Handle variable length sequences
5. Stack multiple RNN layers