long-term short-term memory networks

Long-term short-term memory networks (LTSM) are used on time series data. LTSM have special memory nodes that can be triggered to remember the state of their input based on the values of other parts of a neural network. This can be used to model effects such as systems that have long phases of similar behaviour, with occasional changes to other kinds of behaviour, or to recall important events for long periods.

Defined on page 332

Used on page 332

Also known as long-term short-term network, LTSM