🧠 1. Recurrent Neural Networks (RNNs)

✅ Why Use RNNs?

• RNNs are designed for sequential data where order matters.
• Unlike CNNs, which are good at local spatial features, RNNs model dependencies over time.
• Perfect for:
  • Natural language processing
  • Time-series forecasting
  • Speech recognition

✅ Problem it Solves:

• Captures temporal dependencies: Previous outputs influence future predictions.
• Models variable-length input sequences.

⚙️ How Does an RNN Solve This?

• RNNs share weights across time steps.
• Hidden state hₜ acts as memory: h_t = f(W_{xh} x_t + W_{hh} h_{t-1} + b)
• This “loop” allows information to persist.

✅ Modes of RNNs:

• Many-to-One: Sentiment classification.
• One-to-Many: Image captioning.
• Many-to-Many: Machine translation, video classification.

🔥 2. Computing Loss & Backpropagation in RNNs