An input gate is a component of a neural network, particularly in Long Short-Term Memory (LSTM) networks, that determines the extent to which new information is added to the cell state. It plays a crucial role in controlling the flow of information and maintaining the network's ability to learn temporal patterns effectively.