Recurrent Neural Networks
Recurrent Neural Networks (RNNs) are designed for processing sequential data, capable of capturing temporal dependencies.
Contents:
- RNN Principles — Recurrent structure, unrolling through time, vanishing & exploding gradients
- LSTM — Gating mechanisms, long-term memory, forget gate & input gate
- Seq2Seq — Encoder-decoder architecture, teacher forcing
- GRU — Simplified gating mechanism, update gate & reset gate