AI Glossary

Long short-term memory

Also called LSTM

What does “Long short-term memory” mean?

Long short-term memory: A type of recurrent neural network, introduced by Hochreiter and Schmidhuber in 1997, that adds gates controlling what information to keep, update or discard. These gates help it remember useful context over much longer stretches than a basic RNN.

Long short-term memory visual guide: A type of recurrent neural network, introduced by Hochreiter and Schmidhuber in 1997, that adds gates controlling what information to keep, update or discard. These gates help it remember useful context over much longer stretches than a basic RNN.
Visual guide Open full size (new tab)

What is an example of “Long short-term memory”?

LSTMs were the workhorse for speech recognition, handwriting recognition and translation through the 2010s, and still appear in some time-series forecasting.

Keep learning

See also

Recurrent neural networkTermTransformerTermState-space modelTerm

Same topic

Mixture of expertsTermLatent spaceTermMulti-head attentionTermGPTTermOpen-weight modelTerm