What type of recurrent neural network is used in machine learning?

• A. Long short-term memory (LSTM) networks
• B. Transformer architecture
• C. Generative adversarial networks (GANs)
• D. Model assessment

Answer: (A)

Recurrent neural networks process sequences by passing a hidden state from one time step to the next, so they can carry context through the sequence. Among these, long short-term memory networks are designed to remember information over long horizons. They achieve this with memory cells and three gates—forget, input, and output—that control what information to keep, what new information to add, and what to reveal to the next step. This gating helps prevent vanishing and exploding gradients, allowing the model to learn dependencies across many time steps. Training uses backpropagation through time to adjust the network based on sequence errors. This makes LSTM a common choice for sequential tasks like language modeling, speech recognition, and time-series forecasting. By contrast, the transformer relies on attention rather than recurrence, GANs are generative models not designed as recurrent networks, and model assessment isn’t a type of network.