General

---

Recurrent neural network differs from the feedforward networks in the connections of neurons. In an RNN, it is possible that the neurons are connected to other neurons of the view. In addition, it is also possible that connections are switched to a layer in front of it. This connection data is often necessary when data is dependent on temporal factors.

The RN networks can be divided into four different types of RNN types.

1. direct feedback: output of the neuron is used as input signal

2. lateral feedback: the output of one neuron is connected to the input of another neuron of the same layer

3. Indirect Feedback: Connects a neuron of the previous layer to a neuron of the following layer. Wherein the output of the second neuron is used and connects to the previous neuron as an input signal.

4. Complete Connection: Each neuron is connected to every other neuron. It follows an extremely complex system.

Especially in machine learning, an LSTM network is often used, which is also based on direct feedback and therefore an extension of the RNN.