Recurrent Neural Networks (RNNs) are a type of deep learning model that bakes in assumptions about the importance of the sequential nature of the input data. This is especially useful for time series and natural language data.