Recurrent neural networks
Recurrent neural networks (RNNs) are a paradigm for neural network architectures. They allow to model sequential or temporal data.
Solves the problem of vanishing and exploding gradients of previous architectures. Currently it is the prevalent RNN architecture.
Similar to LSTM’s structure, but computationally more efficient. Its adoption is on the rise.
- Simple recurrent networks (SRNs)
- Elman network
- Jordan network