Recurrent Neural Network (RNN)

A class of artificial neural networks where connections between nodes form a directed graph along a temporal sequence, good for sequential data.

How it works

RNNs process sequences one element at a time. At each step, the network receives the current input and a hidden state vector that carries information from all previous steps. The hidden state is updated by combining the current input with the previous hidden state through learned weight matrices. LSTM and GRU variants add gating mechanisms that control what information is retained or forgotten, addressing the vanishing gradient problem that made early RNNs unable to learn long-range dependencies.

Why it matters

RNNs were the dominant architecture for sequential data — text, speech, time series — before Transformers. They powered the first wave of practical neural machine translation and speech recognition systems. While Transformers have largely replaced them for language tasks, RNNs remain relevant for applications where processing must happen strictly in time order, on very long sequences where attention's quadratic cost is prohibitive, or on devices too small to run large Transformer models.