Similarities and differences between traditional neural network and circulating neural network

The similarities and differences are as follows:

1, Similarity: Both are machine learning models for processing data. Both of them need to optimize the parameters of the model through learning process.

2. Difference: The traditional neural network is a static network, in which the flow of information is forward. Circular neural network has a ring structure, and information can circulate in the network, which makes it possible to process sequence data one by one. Traditional neural network is more suitable for processing static data, while cyclic neural network is more suitable for processing dynamic and time series data. Traditional neural networks usually use back propagation algorithm and gradient descent algorithm for learning and training. On this basis, because of its circular structure, the recurrent neural network can use more complex optimization algorithms such as gradient descent method and momentum method to optimize the weight parameters.