Long Short-Term Memory (LSTM) is a term used in artificial intelligence and big data. LSTMs are special types of artificial neural networks that are particularly good at recognising patterns in data series that extend over longer periods of time.
Imagine you want to predict sales for the coming months based on past sales figures. Conventional methods struggle with this because they often only look at current figures. LSTMs work differently: they remember important information from the past and can therefore better predict what will happen next.
Typical areas of application include forecasting demand, speech recognition or automatically recognising fraud patterns in financial data. Long-term memory shows its strengths particularly in tasks where processes or changes over time are important.
An example from everyday life: voice assistants such as Siri or Alexa use LSTM models to correctly understand spoken sentences - because they understand the context of the individual words in a sentence. This makes LSTM a key technology for many modern digital applications.
Deutsch 
English (UK)